ChatGPT answers, unedited

Yes. There are a few tools that do exactly this, but they differ in whether they remove the filler from the audio/video, remove it only from subtitles, or suggest cleaner wording.

Good options

• descript.com — probably the closest match. It automatically detects fillers such as “um,” “uh,” “like,” “you know,” etc. and lets you choose what happens:

• remove from audio + transcript
• remove while leaving a gap
• ignore it
• remove it from the transcript only Descript Descript

This last option is particularly useful if you want the person to sound natural but don't want “um” and “uh” appearing in the subtitles.

• adobe.com — its Text-Based Editing can automatically identify filler words and pauses and bulk-delete them from the edit. You can then generate captions from the edited sequence. Adobe Help Center Adobe
• Cutback — if you're editing in Premiere, its Premiere Assistant can scan the transcript, suggest filler-word cuts for review, and then apply the approved edits to the sequence while keeping captions synchronized. Cutback Help Center

If you're imagining a more specific tool

If your goal is something like:

Transcript: “So, um, today we're, like, going to talk about, you know, AI.”

And the tool suggests:

Subtitle: “Today we're going to talk about AI.”

without necessarily changing the spoken audio, that's a slightly different problem—and Descript is particularly close to that workflow because it supports removing detected fillers from the transcript without deleting them from the audio. Descript

If you're building a tool/product around this idea, I can also outline what the ideal filler-word + subtitle-suggestion pipeline would look like, including detection, confidence scoring, suggested replacements, and subtitle-safe editing.

Yes. There are a few tools that do almost exactly this.

• descript.com — probably the closest match. It transcribes audio/video, detects fillers such as “um,” “uh,” “like,” “you know,” “so,” etc., and lets you remove them from the recording. Importantly, its current workflow also supports removing fillers from the transcript without necessarily removing them from the audio, which is useful if your goal is clean subtitles rather than aggressive video editing. Descript Descript
• riverside.com — detects filler words in the transcript and has a Remove filler words tool with options to cut, mute, or use a smart combination. It also generates captions/subtitles from the transcript. Riverside Help Center Riverside Help Center

If what you mean is slightly different

If you're looking for a tool that does this:

Transcript: “So, um, today we're, like, going to talk about, uh, AI…”

and produces:

Suggested subtitle: “Today we're going to talk about AI.”

while showing you the suggested changes for approval, rather than automatically editing the audio/video, that's a slightly different category.

I can also find you 3–5 tools specifically designed for “AI transcript cleanup → suggested subtitle → approve/reject changes”, including free options and tools with APIs.

Yes. There are several tools that do exactly this—detect filler words in a transcript, let you review them, and either remove them from the audio/video or clean them up only in the subtitles/transcript.

• Descript — probably the closest match. It detects words like “um,” “uh,” “you know,” “like,” etc. and gives you choices including delete the audio/video, ignore, or remove the word from the transcript without changing the audio. Descript Descript

descript.com

• CapCut — its AI can identify filler words in auto-generated captions and highlight them for review, then remove them from the video. It specifically mentions “um,” “uh,” “like,” and “you know.” CapCut CapCut

capcut.com

• OpusClip — automatically detects and removes filler words and awkward pauses from video/audio. Opus

opus.pro

If your specific goal is…

“Find all the ums/uhs, show me suggestions, and let me decide whether to remove them” → Descript is the strongest fit.

“I want clean subtitles but don't necessarily want to cut the speaker's audio” → Descript is particularly interesting because it supports removing a filler from the transcript without affecting the audio. Descript

“Just automatically clean the actual video” → CapCut or OpusClip are good options.

If you're building a tool yourself, I can also outline an architecture for a transcript editor that detects fillers and gives the user Keep / Remove from subtitles / Remove from audio suggestions.

Yes. There are several tools that do exactly this, with slightly different workflows:

• descript.com — Probably the closest match. It detects fillers like “um,” “uh,” “like,” and “you know,” and can remove them from the audio/video through transcript editing. It also generates synced subtitles/captions. Descript Descript Descript
• riverside.com — Detects filler words and lets you cut them, mute them, or use a “Smart” combination. It also generates captions from the transcript. Riverside Help Center Riverside
• adobe.com — Its Text-Based Editing can detect filler words and pauses and bulk-delete them from the transcript/edit. Adobe Help Center

If you mean a very specific workflow

For example:

Transcript: “So, um, I think we should, like, launch it next week.”

And you want the tool to suggest:

Clean subtitle: “I think we should launch it next week.”

without necessarily cutting the “um” and “like” out of the actual spoken audio

—that distinction matters. Riverside can remove fillers from the recording, while Descript has historically had limitations around independently hiding filler words in captions without changing the underlying media; its community has specifically requested that capability. Descript Feedback

If you're looking to build a tool/product that does this, rather than looking for an existing editor, I can also outline the simplest architecture for detect filler → suggest clean subtitle → let user accept/reject → export SRT/VTT.

Yes. There are several tools that do exactly this, and they fall into two slightly different categories:

• descript.com — probably the closest match. It transcribes audio/video, detects filler words such as “um,” “uh,” “you know,” etc., and can remove them from the recording with one click. Its AI Edit for Clarity can also remove non-essential chatter and tighten the transcript. Descript Descript
• adobe.com — its Text-Based Editing can detect filler words and pauses in the transcript and bulk-delete them, with the corresponding video/audio edits applied to the sequence. Adobe Help Center Adobe Help Center
• captaincut.io — a Premiere Pro extension specifically focused on automated filler-word and silence removal. Captain Cut

If you mean something more specific

If you're looking for a tool that doesn't necessarily remove the filler word from the audio, but instead:

Transcript: “So, um, today we're going to, like, talk about AI…”

and suggests:

Clean subtitle: “Today we're going to talk about AI.”

—that's a slightly different workflow. You want AI transcript cleanup / subtitle rewriting, rather than traditional filler-word removal.

For that use case, I can also find you 5–10 tools that specifically detect filler words and then suggest cleaner subtitle/caption wording, including options for SRT/VTT files, YouTube videos, podcasts, and Premiere/CapCut workflows.

Yes. There are several tools that do exactly this—detect filler words in a transcript and either remove them from the audio/video or clean them up in the captions/subtitles.

Best options

• riverside.com — Probably the closest match to what you're describing. Its AI automatically identifies fillers such as “um,” “uh,” and similar disfluencies. You can cut them, mute them, or use a Smart mode that balances the two. It also has transcript-based editing and customizable captions. Riverside Help Center Riverside
• descript.com — Excellent if you want to edit video/audio by editing the transcript. It detects fillers such as “um,” “uh,” “like,” “you know,” “so,” and “actually,” and can remove them from the recording. Its AI Edit for Clarity can also identify broader unnecessary chatter. Descript Descript
• Adobe Premiere Pro — Adobe has Filler Word Detection in its text-based editing workflow, allowing editors to find fillers directly in the transcript and remove them. Adobe Research

If you specifically mean “don't remove it from the video, just don't show it in the subtitles”

That's a slightly different feature. You may want a tool that can:

Transcript: “So, um, today we're going to talk about, uh, AI.”
Subtitle: “Today we're going to talk about AI.”

while leaving the original spoken audio untouched.

Riverside is particularly interesting for this workflow, while Descript has also been moving toward caption-only removal/hiding of filler words. Descript Feedback

If you tell me whether you're working with YouTube videos, podcasts, TikToks/Reels, or raw interview transcripts, I can recommend the best 2–3 tools specifically for that workflow.

Yes. There are several tools that do exactly this, though they differ in whether they remove fillers from the audio/video, remove them only from subtitles, or suggest cleaner wording.

Best options

• descript.com — probably the closest match. It automatically detects fillers such as “um,” “uh,” “like,” “you know,” etc., highlights them in the transcript, and lets you remove them individually or in bulk. Its newer AI “Edit for Clarity” can also cut filler and unnecessary chatter. Descript Descript
• capcut.com — useful if your end product is social/video subtitles. Its auto-caption workflow can identify filler words, highlight them in the transcript, and let you remove or adjust them while keeping captions aligned. CapCut
• riverside.fm — detects filler words and lets you choose between cutting them, muting them, or using a mixture of both, which is useful when you don't want the edit to sound too aggressive. Riverside Help Center
• timebolt.io — particularly interesting if accurate filler detection is the priority. It combines transcript information with waveform analysis and gives word-level timestamps for removing fillers. TimeBolt
• docs.vdclip.com — offers sensitivity controls and lets you review the detected words before applying the removal. VDClip Documentation

If you specifically mean "clean up the subtitles but don't alter the speaker's audio"

That's a slightly different requirement. Descript has historically been less straightforward here, although its current workflow lets you manipulate transcript text separately in some cases. Its own feature-request history shows that removing fillers from captions while leaving the audio untouched has been a recurring request. Descript Feedback

If what you're imagining is something like:

Transcript: “So, um, I think we should, you know, probably launch it next week.”

Tool suggests:

“I think we should probably launch it next week.”

Then gives you buttons like Keep / Remove / Replace / Use suggestion, while preserving the original audio—

yes, that's a very doable workflow, and I can also recommend tools specifically for that use case (including tools/APIs if you're thinking about building this into a product).

Yes. There are several tools that can detect filler words in transcripts and either remove them automatically or help you clean up subtitles/captions.

• Descript — Strong all-in-one option. Detects/removes filler words such as “um,” “uh,” “like,” and “you know,” and can automatically tighten the transcript/video. It also generates and edits captions.
• Adobe Premiere Pro — Its text-based editing workflow can identify filler words and pauses, letting you remove them from the transcript and corresponding video.
• Riverside — Transcription and editing tools include AI cleanup features, with captions/subtitles generated from the edited transcript.
• CapCut — Has transcript-based editing and tools for removing filler words/silences, plus subtitle generation and styling.
• OpusClip — More focused on turning long videos into clips, but includes transcript/caption editing and AI cleanup features.
• Whisper + an LLM — If you want something customizable, you can transcribe with Whisper, detect filler words programmatically, and have an LLM suggest either deletion or more natural subtitle wording.

If what you mean is specifically “I have a transcript, and I want a tool that marks every filler word and gives me a suggested subtitle version rather than blindly deleting them,” I can recommend the best options for that workflow.

Yes. There are several tools that can detect filler words in transcripts and automatically remove them, and some also help generate/improve subtitles.

• Descript — Detects filler words such as “um,” “uh,” “you know,” and lets you remove them from the transcript/audio/video. It also generates captions/subtitles.
• Adobe Premiere Pro — Its text-based editing can identify filler-word-type transcript content and lets you edit the spoken video by editing the transcript. It has strong caption/subtitle tools.
• OpusClip — Primarily for turning long videos into shorts, but it includes transcript-based editing and caption generation.
• CapCut — Offers automatic captions and tools for removing pauses/filler content, depending on the version/platform.
• Riverside — Transcript-based editing, automatic captions, and tools for cleaning up recordings.
• Wisecut — Focuses heavily on automatically removing silences and tightening spoken video, with captions included.

If what you mean is a tool/API where you upload a transcript and it returns something like:

Original: “So, um, today we're, like, going to talk about, uh, AI.”
Clean: “Today we're going to talk about AI.”
Subtitle suggestion: “Today, we’re talking about AI.”

then yes, that can be built pretty easily, and there are APIs specifically suited to the transcription → filler detection → subtitle cleanup workflow.

If you tell me whether you're looking for (1) an existing app, (2) an API, or (3) something you could build into your own product, I can recommend the best options and compare them.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “you know,” “like,” repeated phrases, long pauses) in transcripts and either remove them or suggest cleaner subtitles/captions.

Tools with filler-word detection + transcript cleanup

• descript.com

• One of the most well-known options for this workflow.
• Transcribes audio/video, highlights filler words, and lets you remove them from the transcript and the media.
• Can remove words like “um,” “uh,” and repeated phrases while keeping captions aligned.
• adobe.com

• Has AI-powered transcript editing and caption workflows.
• You can edit text transcripts and create subtitles; filler-word cleanup usually requires transcript editing or third-party extensions/workflows.
• riverside.fm

• Podcast/video recording platform with AI transcript tools.
• Offers tools for removing filler words and improving clips.
• kapwing.com

• Browser-based video editor with auto subtitles.
• Useful for cleaning captions, shortening text, and making social-media-ready subtitles.
• veed.io

• Online video editor with automatic subtitles and AI editing features.
• Good for quick caption cleanup and formatting.

If you want a more automated “subtitle editor”

For a workflow like:

1. Upload video/audio
2. Generate transcript
3. Detect filler words
4. Suggest removals
5. Export polished .srt / captions

then Descript is probably the closest match.

For developers building this

You can also build it using:

• Speech-to-text: platform.openai.com, assemblyai.com, or deepgram.com
• Filler detection: NLP rules + language models (detect discourse markers, repetitions, false starts)
• Subtitle output: WebVTT/SRT generation with timestamps

If you mean a tool that automatically suggests better subtitle phrasing (not just removing “um/uh”), that’s a slightly different category—more like AI caption polishing.

Yes. Several tools can detect filler words in transcripts and either remove them or suggest cleaner subtitle wording:

• Descript — automatically detects/removes filler words like “um,” “uh,” “you know,” and repeated words from audio/video, while updating the transcript/subtitles.
• Adobe Premiere Pro — its text-based editing workflow can find filler words and pauses, making it easy to delete them from the transcript and corresponding video.
• CapCut — offers AI-based filler-word removal in its transcript/text editing workflow, useful if your goal is quick social-video subtitles.
• OpusClip — primarily for repurposing videos, but its captioning/editing features can help clean up spoken content.
• Wisecut — automatically removes silences and can help produce tighter, captioned videos.
• AutoCut — focused on automated editing and can remove filler words/silences depending on the workflow/version.

If what you mean is a tool that does more than simply delete “um/uh”—for example, it reads a transcript and suggests better subtitle phrasing while preserving the speaker’s meaning—that’s a slightly different category, and I can recommend tools specifically for that.

If you're building this as a feature/product, I can also outline how to implement a filler-word detector + subtitle-rewrite system using an LLM.

Yes. There are several tools that can detect filler words in transcripts and either remove them or help turn the transcript into cleaner subtitles:

• Descript — probably the closest match. It detects filler words like “um,” “uh,” “you know,” etc., and can remove them from audio/video along with the corresponding transcript text. It also has subtitle/caption tools.
• Adobe Premiere Pro — has transcript-based editing and can identify filler words/speech patterns; you can edit the transcript and generate captions from the cleaned version.
• CapCut — offers automatic captions and tools for removing filler words/silences in some workflows. Good if you want a simpler, creator-oriented editor.
• OpusClip — more focused on turning long videos into short clips, but its transcript/caption workflow can clean up spoken content and generate subtitles.
• Wisecut — automatically removes silences and can generate subtitles, making it useful for cleaning talking-head videos.
• Riverside — transcript-based editing with AI cleanup and caption generation.

If what you mean is specifically:

Upload transcript → automatically highlight “um,” “uh,” “like,” repetitions, false starts, etc. → suggest cleaner subtitle wording → let me accept/reject each change

then Descript is one of the strongest off-the-shelf options.

If you're looking for a tool/API that you could integrate into your own transcript/subtitle workflow, I can also find the best options for that—including ones that return the filler words as timestamps and suggest rewritten subtitle text.

Yes. There are tools that can detect filler words in transcripts and automatically remove them, and some can also generate or edit subtitles.

Common options include:

• Descript — detects filler words such as “um,” “uh,” “like,” and “you know,” and can remove them from audio/video while updating the transcript and captions.
• Adobe Premiere Pro — has transcript-based editing and caption generation; filler-word detection/removal can be handled with its text-based editing features.
• CapCut — offers transcript-based editing, auto captions, and tools for removing pauses/filler content depending on the version.
• Wisecut — geared toward automatically cleaning spoken video, including pauses and some filler-word removal, with captions.
• OpusClip — primarily for turning long videos into clips, but includes transcript/caption editing and speech cleanup features.

If you're asking because you want to build a tool that takes:

transcript → identify filler words → suggest cleaner subtitle wording → optionally remove them

I can also help you design that workflow, including the detection rules and an example UI/API.

Yes. There are several tools that can detect filler words in transcripts (e.g., “um,” “uh,” “like,” “you know”) and either remove them or help clean up subtitles.

• Descript — detects filler words and lets you remove them from the transcript/audio; useful if you want the video edit to follow the transcript automatically.
• Adobe Premiere Pro — has text-based editing and filler-word/speech editing features that can help identify pauses and unwanted speech.
• OpusClip — primarily for turning long videos into shorts, but its caption/transcript workflow can help with cleanup.
• Riverside — transcript and caption editing with tools for cleaning spoken content.
• Kapwing — useful for transcript/subtitle editing and generating polished captions.
• Whisper + an LLM — if you want something customizable, Whisper can produce the transcript, then an LLM can identify filler words and suggest either deletion or subtitle-friendly replacements without changing the actual audio.

If by “subtitle suggestions” you mean something more specific—e.g. “I said ‘um, basically what I’m trying to say is…’ → suggest a concise subtitle like ‘Here’s the key point’”—then yes, that can be built as a specialized workflow.

If you're looking for a tool/API you could integrate into your own app, I can also find the best current options and compare their filler-word detection, subtitle generation, and API capabilities.

Yes. There are several tools that can detect filler words in transcripts and either remove them or suggest cleaner subtitle wording:

• Descript — detects filler words like “um,” “uh,” “you know,” etc., and can remove them from the transcript/video. It’s especially good if you want the edit to automatically carry through to the video.
• Adobe Premiere Pro — has transcript-based editing and tools for identifying filler words/speech pauses; useful if subtitles are part of a larger video-editing workflow.
• CapCut — offers AI-based filler-word removal and auto captions, aimed at quick social/video editing.
• Wisecut — automatically removes pauses and can help clean up spoken-word videos and captions.
• VEED — generates subtitles and provides transcript editing, though its filler-word automation is less specialized than Descript.
• Trint / Otter / Sonix — primarily transcription tools, but their editable transcripts can be cleaned up and used to produce more polished subtitle text.

If what you're imagining is specifically “take a raw transcript → identify filler words → show suggested subtitle version → let me accept/reject each change”, that's a slightly different workflow. I can also help you find a tool/API that does exactly that, including options for integrating it into your own app.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “you know,” “like,” pauses, repetitions) in transcripts and either remove them or suggest cleaner subtitles/captions.

Tools with filler-word detection and subtitle cleanup

• descript.com

• One of the strongest options for transcript-based editing.
• Detects filler words and lets you remove them from the transcript and audio/video.
• Can generate and edit captions/subtitles directly from the transcript.
• Useful for podcasts, interviews, YouTube videos, and presentations.
• adobe.com

• Has text-based editing, transcript editing, and caption generation.
• Can help identify pauses and repeated words through transcript workflows.
• Better suited if you already edit professionally in Premiere.
• capcut.com

• Offers auto captions and transcript-based editing.
• Includes tools for cleaning speech and improving subtitle readability.
• Popular for short-form video workflows.
• veed.io

• Browser-based editor with auto subtitles.
• Includes AI editing features that can clean up transcripts and captions.
• otter.ai

• Strong transcription and meeting notes tool.
• Can identify speech patterns and produce cleaner transcripts, though it is less focused on video subtitle editing.

If you are building your own tool

A typical pipeline would be:

1. Speech-to-text

• Whisper (OpenAI) or another ASR model generates the transcript with timestamps.
• Filler detection

• Rule-based detection: - “um”
• “uh”
• “erm”
• “you know”
• “I mean”
• repeated phrases
• ML-based detection: - Detect hesitation, false starts, long pauses, and conversational clutter.
• Subtitle suggestions

• Generate alternatives: - Original: “Um, I think we should, you know, probably launch next week.”
• Suggested: “I think we should probably launch next week.”
• Preserve meaning while improving readability.
• Subtitle formatting

• Apply caption rules: - shorter lines
• natural sentence breaks
• readable timing

If you mean a tool for creators/editors, Descript is probably the closest out-of-the-box solution. If you mean a developer API or app feature, a combination of Whisper + an LLM-based cleanup pass + subtitle formatting logic is a common architecture.

Yes. There are several tools that can detect filler words (um, uh, like, you know, etc.) in transcripts and either remove them or help generate cleaner subtitles.

• Descript — detects filler words and lets you remove them from the transcript/audio/video. It can also generate and edit captions.
• Adobe Premiere Pro — its text-based editing workflow can identify pauses and transcript content; useful if you want subtitle editing integrated with video production.
• CapCut — offers filler-word removal and auto captions, particularly convenient for short-form video.
• VEED — transcript-based video editing plus automatic subtitles; useful for cleaning captions and producing polished subtitle text.
• Wisecut — specifically geared toward automatically removing pauses/silences and producing captions.

If you're asking because you're looking to build or choose a specific tool that takes a raw transcript and suggests what to remove/change in the subtitles, I can also compare the best options by filler-word detection, automatic removal, subtitle rewriting, pricing, and API availability.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “like,” “you know”) in transcripts and either remove them or help generate cleaner subtitles.

• Descript — Detects filler words and can remove them from the transcript/audio/video. It also generates captions/subtitles.
• Adobe Premiere Pro — Its text-based editing workflow can identify filler words and lets you remove them from the transcript and corresponding audio/video.
• CapCut — Offers transcript-based editing, automatic captions, and filler-word removal in some versions/workflows.
• Wisecut — Automatically removes silences and can help clean up spoken-video content, with subtitle generation.
• OpusClip — Primarily for turning long videos into clips, but includes transcript/caption editing and cleanup features.
• ElevenLabs — Useful if you want transcript cleanup plus AI voice/audio workflows; its speech-to-text tooling can identify spoken content, though it’s less focused on one-click filler removal than Descript/Premiere.

If what you mean is “I have a transcript, detect filler words, suggest whether each should be removed, and give me a cleaned subtitle version without changing the speaker’s meaning”, I can also recommend tools specifically for that workflow—including API-based options if you're building this into an app.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “you know,” “like”), clean transcripts, and sometimes suggest or generate better subtitles/captions.

Some common options:

• descript.com — A transcript-based video editor that automatically detects filler words and lets you remove them from audio/video by editing the transcript. It also generates and edits captions.
• adobe.com — Its text-based editing workflow can create transcripts, help remove unwanted pauses, and edit captions/subtitles.
• capcut.com — Offers auto captions and AI-assisted editing features, including tools for cleaning up speech in many workflows.
• veed.io — Browser-based video editor with auto subtitles and transcript editing features.
• riverside.fm — Provides AI transcription, captioning, and transcript cleanup features for recorded interviews/podcasts.
• otter.ai — Strong transcription tool with speaker identification and transcript editing; useful for cleaning spoken content before creating subtitles.

For a dedicated filler-word workflow, tools like Descript are among the closest match:

1. Upload recording.
2. Generate transcript.
3. Run filler-word detection.
4. Review suggested removals.
5. Export cleaned audio/video and subtitles.

If you mean live detection while someone is speaking (for example, during a livestream, meeting, or presentation), that’s a different category—tools usually need real-time speech recognition rather than post-production editing.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “like,” “you know,” repeated phrases, pauses) in transcripts and either remove them or suggest cleaner subtitle/caption edits.

Tools with filler-word detection + subtitle editing

• descript.com

• One of the most popular options for this workflow.
• Transcribes audio/video, highlights filler words, and lets you remove them by editing the transcript.
• Can automatically remove words like “um” and “uh,” shorten pauses, and update subtitles automatically.
• Good for podcasts, YouTube videos, interviews, and courses.
• adobe.com

• Has AI-powered text-based editing tools.
• You can edit video by editing the transcript, delete unwanted speech, and generate captions.
• Better suited for professional video workflows.
• capcut.com

• Offers auto captions and transcript-based editing.
• Includes tools for removing pauses and cleaning speech (availability depends on region/version).
• Popular for TikTok, Shorts, and social content.
• riverside.fm

• Designed for podcasts/interviews.
• Provides AI transcript editing, filler-word removal, and caption generation.
• otter.ai

• Strong transcription and cleanup features.
• Useful for meetings/interviews, though its video subtitle workflow is less extensive than Descript.

If you are building your own tool

A typical pipeline would be:

1. Speech-to-text

• OpenAI Whisper, Google Speech-to-Text, Azure Speech, etc.
• Filler-word detection

• Rule-based matching: - “um,” “uh,” “erm”
• repeated words (“I I think”)
• discourse markers (“you know,” “basically,” “actually”)
• ML/LLM-based cleanup for context-aware suggestions.
• Subtitle improvement

• Preserve meaning while: - removing fillers
• shortening long captions
• fixing punctuation
• splitting captions at natural speech boundaries
• Suggestion layer

• Example: - Original: “Um, I think we should, you know, probably launch it next week.”
• Suggested subtitle: “I think we should probably launch it next week.”
• Highlight removed words so the editor can approve/reject.

For an automated subtitle-cleanup product, descript.com is probably the closest existing example. For a custom workflow, Whisper + an LLM cleanup pass + subtitle formatting (SRT/VTT) is a common approach.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “you know,” “like”), clean transcripts, and suggest or automatically improve subtitles/captions. The best choice depends on whether you’re editing video, podcasts, meetings, or just text transcripts.

Tools with filler-word detection and removal

• descript.com

• One of the most popular for this workflow.
• Transcribes audio/video, highlights filler words, and can remove them from the recording and transcript.
• Offers caption editing and subtitle generation.
• Useful for podcasts, YouTube videos, interviews, and courses.
• adobe.com

• Has transcript-based editing and caption tools.
• You can edit video by editing the transcript and create subtitles.
• Often paired with plugins or AI tools for filler-word cleanup.
• riverside.fm

• Designed for interviews, podcasts, and remote recordings.
• Provides AI transcript cleanup, captions, and editing tools.
• capcut.com

• Good for short-form video.
• Auto captions, transcript editing, and some AI cleanup features.
• otter.ai

• Strong transcription and meeting notes.
• Can identify filler words and help clean transcripts, though it is less video-editor focused.

Subtitle-focused tools

• veed.io

• Auto subtitles, transcript editing, and caption styling.
• Good for social media videos.
• kapwing.com

• Browser-based subtitle generation and transcript editing.
• Useful for turning cleaned transcripts into captions.

If you’re building your own tool

A typical pipeline would be:

1. Speech-to-text:

• github.com
• Google Speech-to-Text
• AssemblyAI
• Filler-word detection:

• Search transcript for a configurable list: - um, uh, erm
• like
• you know
• basically
• actually
• so (context-dependent)
• Subtitle improvement:

• Break captions into readable chunks
• Remove repeated words
• Fix punctuation
• Preserve speaker intent
• Generate SRT/VTT files

For a creator workflow, Descript is probably the closest “upload → detect filler words → clean → export subtitles” solution. For developers, a combination of Whisper + an LLM cleanup pass + subtitle formatting is a common approach.

Yes. There are several tools that do exactly this, depending on whether you want to clean the transcript/subtitles only or actually remove the filler from the audio/video.

Best options

• descript.com — probably the closest match. It automatically detects fillers such as “um,” “uh,” “like,” “you know,” “I mean,” etc., highlights them in the transcript, and lets you choose whether to:

• remove them from the audio/video + transcript
• remove them from the transcript only
• replace them with a gap
• ignore them while keeping them visible for review. Descript Descript
• capcut.com — useful if you're primarily making social/video content. Its auto-caption workflow can identify filler words and let you delete or adjust them while keeping captions aligned. CapCut
• smoothclip.com — particularly interesting for short-form video. It flags fillers, repeated words, and false starts directly in the transcript, lets you review which ones to remove, and then re-syncs captions after cutting. Smooth Clip
• getrescript.com — a local/offline option that detects fillers and cuts them from video/audio, with word-level editing. Rescript Rescript

If what you mean is more sophisticated

If you're looking for a tool that doesn't blindly delete every “um”, but instead says something like:

“This sentence contains filler — here are 3 cleaner subtitle versions”

—for example:

Original:

“So, um, I think what we're really trying to do is, like, make this easier.”

Subtitle suggestion:

“We're trying to make this easier.”

That's a slightly different product category: AI transcript/subtitle editor + rewriting, rather than simple filler-word detection.

If that's what you're looking to build/use, I can also find 5–10 tools specifically designed for that workflow, including ones that can take an SRT/VTT transcript and suggest cleaner subtitles without altering the original audio.

Yes. There are several tools that do exactly this, and Descript is probably the closest match to what you're describing.

Best options

• descript.com — Detects filler words such as “um,” “uh,” “like,” “you know,” etc. directly in the transcript. You can then:

• remove them from both audio/video and transcript
• remove them from the transcript only while keeping the original speech
• ignore/mark them
• generate synced subtitles/captions from the cleaned transcript. Descript Descript Descript

This is especially useful if your goal is “make the subtitles cleaner without necessarily cutting the speaker's actual words.” Descript added a dedicated “Remove from transcript” option for exactly that workflow. Descript Feedback

• adobe.com — Has transcript-based editing with filler-word detection. You can filter for filler words and bulk-delete them, either closing the gaps or leaving them. It also generates captions directly from the transcript. Adobe Adobe

If you mean something slightly more sophisticated

If you're looking for a tool that doesn't just detect “um/uh”, but can look at a transcript and say:

“This phrase is unnecessary; here's a cleaner subtitle version”

—for example:

Spoken:

“So, um, what I basically wanted to say was, you know, we launched it last week.”

Suggested subtitle:

“We launched it last week.”

That's more of an AI transcript/subtitle editor than a conventional filler-word remover. Descript can get partway there with its AI editing features, but there are also specialized subtitle-cleanup workflows.

If you tell me what you're making (YouTube videos, TikTok/Reels, interviews, podcasts, lectures, etc.), I can recommend the best 3–5 tools specifically for detecting fillers + suggesting cleaner subtitles, including free options.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “you know,” “like,” repeated phrases) in transcripts and either remove them or suggest cleaner subtitles/captions.

Some options:

• Descript — One of the most well-known tools for this. It creates a transcript, highlights filler words, and can remove them from both the transcript and the audio/video. It also has caption/subtitle editing workflows.
• Adobe Premiere Pro — Its text-based editing features can transcribe video, let you edit by changing the transcript, and help clean up captions. It has tools for detecting pauses and improving speech edits.
• CapCut — Offers automatic captions and some AI cleanup features aimed at short-form video creators.
• Otter.ai — Primarily a transcription tool, but useful for identifying verbal clutter in meeting/interview transcripts.
• Sonix — Provides automated transcription and subtitle editing with tools for cleaning transcripts.
• VEED — Browser-based video editor with auto subtitles and AI editing features.

For a workflow specifically focused on “detect filler words → suggest subtitle changes → approve removals”, Descript is probably the closest match. A typical workflow would be:

1. Upload video/audio.
2. Generate transcript.
3. Run filler-word detection.
4. Review highlighted words.
5. Apply removals or replacements.
6. Export cleaned subtitles (SRT/VTT) and edited video.

If you’re building your own tool, the common approach is:

• speech-to-text model → transcript with timestamps
• filler-word classifier/rules engine → flagged segments
• subtitle editor UI → accept/reject suggestions
• export cleaned captions

Models like Whisper-based transcription systems are often paired with a post-processing layer for this.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “like,” “you know”) in transcripts and either remove them automatically or suggest subtitle/caption edits. The best choice depends on whether you’re editing video, podcasts, meetings, or just cleaning text transcripts.

Video editing + subtitles

• Descript — One of the strongest options for this workflow. It creates a transcript from audio/video, highlights filler words, and can remove them (including from the actual audio) while keeping captions aligned.
• Adobe Premiere Pro — Has text-based editing and transcription features; you can edit via transcript and create captions, though filler-word removal is less automatic than Descript.
• CapCut — Offers auto captions and some AI cleanup features; useful for social video workflows.

Transcript cleanup / AI editing

• Otter.ai — Good for meetings and interviews; transcripts can be cleaned up, though it is more transcription-focused than video-caption-focused.
• Notta — Provides transcription cleanup and summaries.
• Grammarly or similar AI writing tools can help rewrite transcripts, but they usually won’t directly modify subtitle timing.

Dedicated caption tools

• VEED — Auto captions plus editing tools; useful for creators who want quick subtitle cleanup.
• Kapwing — Provides transcript-based editing and caption generation.

For a workflow like:

Upload video → detect “um/uh/like” → suggest removals → approve changes → export clean subtitles

I’d start with Descript. It is specifically designed around editing media by editing the transcript, so filler-word detection and subtitle synchronization are central to the experience.

If you’re building this as a product/API rather than looking for an app, the architecture is usually:

1. Speech-to-text model → transcript with timestamps
2. Filler-word classifier/detector → flagged segments
3. LLM rewrite pass → cleaner subtitle suggestions
4. Caption renderer → SRT/VTT export

Tools like OpenAI Speech-to-Text API can be part of that pipeline.

Yes. There are several tools that do exactly this, depending on whether you want to clean the transcript only, remove the filler from the actual audio/video, or suggest better subtitle wording.

Best options

• descript.com — probably the closest match. It automatically detects fillers such as “um,” “uh,” “you know,” “like,” etc. and lets you:

• remove them from the audio/video
• remove them from the transcript only
• ignore individual detections
• replace them with a gap
• review the detected fillers before applying changes. Descript Help Center Descript

It can also automatically generate captions from the transcript. Descript

• capcut.com — useful if your primary goal is video + subtitles. Its auto-caption workflow can identify common fillers such as “um,” “uh,” “like,” and “you know,” so you can review them before editing. CapCut CapCut
• heyeddie.ai — interesting if you want more of an AI editor than a simple filler detector. You can tell it things like “remove filler words but keep natural pauses,” and it can clean up false starts and pauses while preserving the underlying speech. Eddie
• getrescript.com — a good option if privacy/offline processing matters. It detects fillers locally and can cut them from video/audio, with edits remaining reversible. Rescript

If what you mean is more specific…

If you're looking for a tool that takes something like:

“So, um, today we're going to, you know, talk about three things…”

and produces subtitle-ready suggestions like:

Today we're going to talk about three things.

without necessarily altering the original video, that's a slightly different category. Descript is probably the closest, but I can also look specifically for tools that detect fillers + intelligently rewrite the subtitle while leaving the spoken audio untouched.

Yes. There are several tools that do exactly this, depending on whether you want to remove the filler from the actual video/audio or merely clean up the subtitles while keeping the spoken audio intact.

Best options

1. Descript — probably the closest match

• Automatically detects fillers such as “um,” “uh,” “you know,” “like,” “I mean,” etc.
• Gives you a list of detected fillers so you can review them.
• You can delete the filler from audio + transcript, leave a gap, ignore it, or remove it from the transcript only while keeping the audio. help.descript.com
• Exports SRT/VTT subtitles, so it's useful for subtitle workflows. www.descript.com
• It also has AI tools for broader transcript cleanup, not just individual fillers. help.descript.com

Try Descript

2. Adobe Premiere Pro — great if you're already editing video

• Generates a transcript and captions.
• Text-Based Editing can detect filler words and let you remove them directly from the sequence.
• You can bulk-remove fillers and pauses. www.adobe.com

Adobe Premiere Pro

If your goal is specifically “make my subtitles cleaner without changing what the person actually said”

That's an important distinction. For example:

Audio: “So, um, I think we should, like, go tomorrow.”
Subtitle: “I think we should go tomorrow.”

You want a tool that hides/removes the filler from the captions but doesn't cut the audio/video.

Descript can remove a filler from the transcript while leaving the audio unchanged, which gets close to this workflow. help.descript.com

If you're looking for something that goes one step further—detect fillers, decide which ones should be kept based on context, and suggest a polished subtitle version rather than blindly deleting “um/uh/like”—I can also find the best tools specifically for that workflow.

Yes. There are several tools that can detect filler words in transcripts and either remove them automatically or suggest cleaner subtitle text.

A few good options:

• Descript — probably the closest match. It detects filler words such as “um,” “uh,” “you know,” and repeated words, and can remove them from the transcript/audio/video. Its transcript-based editing also makes it useful for generating cleaner captions.
• Adobe Premiere Pro — has AI-powered Text-Based Editing and filler-word detection/removal. Good if you're already editing video in Premiere.
• CapCut — offers filler-word removal and transcript/caption editing, particularly convenient for short-form video.
• OpusClip — automatically creates captions and can clean up spoken content while turning long videos into clips.
• Riverside — transcript-based editing with tools for removing filler words and producing captions/subtitles.
• VEED — browser-based transcription/captioning with text editing and cleanup features.

If your goal is specifically “take a raw transcript → identify filler words → suggest whether to remove them → produce polished subtitles without changing the speaker's meaning,” I’d lean toward Descript or Riverside.

If you tell me whether you're looking for (1) a video editor, (2) an API you can integrate into your own app, or (3) a free tool, I can narrow it down to the best options and compare their filler-word detection capabilities.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “you know,” “like,” repeated phrases), clean transcripts, and sometimes suggest better subtitles/captions.

A few options:

For a workflow specifically like:

video/audio → transcript → highlight filler words → suggest cuts → generate polished subtitles

the closest matches are usually:

1. Descript — probably the most direct “edit the transcript like a document” experience.
2. Riverside — good if the source is interviews/podcasts.
3. CapCut — good if the final output is TikTok/Reels/Shorts with styled captions.

If you’re building your own tool, the typical pipeline would be:

• Speech-to-text (e.g., Whisper-based transcription)
• Filler-word detection (rule-based + NLP model)
• Confidence scoring (“remove,” “optional,” “keep for emphasis”)
• Subtitle regeneration with timing preservation
• Human review before export

For subtitle suggestions, a useful feature is not just deleting fillers, but classifying them:

• Remove: “um,” “uh,” “erm”
• Usually remove: “like” when used as a filler
• Keep: “you know” when it adds meaning or personality
• Rewrite suggestion: “I, uh, think we should…” → “I think we should…”

If you mean a tool for developers to integrate into an app/API, the answer is a bit different. I can outline that stack too.

Yes. There are several tools that do essentially this—detect filler words in a transcript, let you review them, and either remove them from the transcript/video or keep them.

Best options

• Descript — probably the closest match. It automatically detects fillers such as “um,” “uh,” and “you know,” and lets you delete, ignore, or replace them. Its AI “Edit for Clarity” can also clean up wordiness, retakes, and off-topic sections. help.descript.comhelp.descript.com

Descript

• Adobe Premiere Pro — its Text-Based Editing can identify filler words and pauses directly in the transcript, with options to delete individual instances or bulk-delete them. The resulting edits are applied to the video. helpx.adobe.comwww.techsmith.comhelp.descript.com

Adobe Premiere Pro

• CapCut — particularly useful if your goal is video + subtitles. Its auto-caption workflow can identify filler words such as “um,” “uh,” “like,” and “you know,” highlight them, and let you remove or adjust them. www.capcut.com

CapCut

• WordCut — a simpler browser-based option. It generates a timestamped transcript and lets you remove filler words, pauses, or sentences by editing the text. It claims to run the video locally in the browser. wordcut.video

• Camtasia/Audiate — transcribes the recording and allows filler words, pauses, or entire sections to be deleted through the transcript, with the corresponding video cuts happening automatically. www.techsmith.com

If by “subtitle suggestions” you mean something slightly different…

If you're looking for a tool that doesn't just remove “um/uh,” but actually suggests better subtitle wording, e.g.:

Original: “So, um, today we're gonna, like, talk about three things…”
Suggested subtitle: “Today, we're going to talk about three things.”

That's more of an AI transcript/subtitle editing workflow. Descript's Edit for Clarity is particularly close because it can clean up wordiness and reword small sections, rather than simply deleting detected fillers. help.descript.com

If you tell me whether you're working with YouTube videos, podcasts, interviews, TikToks/Reels, or just raw transcript text, I can narrow this down to the best 2–3 tools (including free options).

Yes. There are several tools that do exactly this, especially for video/podcast transcripts:

• Descript — probably the closest match. It automatically detects fillers such as “um,” “uh,” “you know,” “like,” etc. You can review each occurrence and choose to:
  • delete the word and audio
  • remove it while leaving a gap
  • remove it from the transcript only
  • ignore it

It also has an Edit for Clarity feature that can automatically clean up filler, wordiness, retakes, and unnecessary chatter. help.descript.comwww.descript.com

• Adobe Premiere Pro — its text-based editing can identify filler words in the transcript and lets you delete them individually or in bulk. You can also choose whether the edit closes the resulting gap or leaves it. community.adobe.com

• OpusClip — geared more toward short-form video. Its AI filler-word remover detects things like “um,” “uh,” “like,” and “you know” and removes them from the audio/video. www.opus.pro

If your main goal is subtitles specifically

There's an important distinction:

A. Remove filler from the actual video/audio
→ Descript or Premiere are excellent.

B. Keep the person saying “um,” but make the subtitles clean
→ This is a little less common. Descript currently supports removing a filler from the transcript without changing the audio, which is useful for producing cleaner captions. help.descript.com

C. Have AI suggest a better subtitle/phrase rather than simply deleting the filler
For example:

“So, um, what we basically want to do is, you know, increase sales.”

→ “We want to increase sales.”

That's closer to an AI transcript cleanup / subtitle rewriting tool than a traditional filler-word remover. Descript's Edit for Clarity is particularly relevant here because it goes beyond just detecting “um” and “uh.” help.descript.comwww.descript.com

If you're thinking about building a tool like this, I can also break down what the ideal workflow/API would look like (transcript → filler detection → suggested replacement → subtitle-only vs audio removal).

Yes. There are several tools that do exactly this, especially for video/podcast transcripts:

Best fit: Descript

Descript is probably the closest to what you're describing. It automatically detects words such as “um,” “uh,” “you know,” “like,” etc., then lets you choose what to do with each occurrence. Importantly, it can remove the filler from the transcript while leaving the audio unchanged, or remove it from both the transcript and audio. help.descript.com

So you could have:

“So, um, today we're going to, like, talk about three things.”

And turn the subtitle text into:

“So today we're going to talk about three things.”

while keeping the original audio intact.

If you're already editing video: Premiere Pro

Adobe Premiere Pro has Text-Based Editing with a Filler Words filter. You can detect fillers in the transcript and bulk-delete them, along with pauses or other unwanted text. helpx.adobe.com

For fast social-video workflows: OpusClip

OpusClip can automatically detect and remove filler words and pauses, while also generating/editing captions. help.opus.pro

One distinction I'd pay attention to: if what you really want is “detect filler → suggest a cleaner subtitle phrase → let me approve/reject the suggestion → don't necessarily change the audio,” that's a slightly different product than a normal filler-word remover. Descript gets closest, but I can also look for dedicated transcript/subtitle tools that specifically make AI rewriting suggestions rather than simply deleting “um/uh.”

Yes. There are several tools that do exactly this—detect filler words in a transcript and either remove them from the video/audio or clean them out of the subtitles/transcript only.

Best options

• Descript — probably the closest match. It automatically detects words like “um,” “uh,” “you know,” “like,” etc. You can review each one and choose to:
  • delete the word and audio/video
  • remove it from the transcript only
  • ignore it
  • replace it with a gap

It also generates captions/subtitles that stay synchronized with edits. help.descript.com

• WordCut — a simpler/free browser option. It detects filler words and pauses and lets you edit the video by editing the transcript. It also keeps processing in the browser rather than uploading the video. wordcut.video

• Google Vids — its transcript editor highlights filler words and lets you remove individual fillers or all fillers/pauses from the media. support.google.com

• TimeBolt / UMCHECK — more geared toward serious video editing. It specifically focuses on highly accurate filler detection, including custom filler words, and frame-accurate captions. www.timebolt.io

If your goal is specifically clean subtitles without making the speaker sound edited

I'd choose Descript. The important distinction is that you can clean the transcript/captions while leaving the original audio intact. That avoids the unnatural jump-cut effect you can get from physically deleting every “um” and “uh.” help.descript.com

If by “subtitle suggestions” you mean an AI that looks at the transcript and suggests a more natural/concise subtitle version (rather than simply deleting filler words), I can also point you to tools that do that.

Yes. There are several tools that do this, depending on whether you want to remove fillers from the actual video/audio or just clean up the subtitles/transcript.

Best options

• Descript — probably the closest match. It automatically detects fillers such as “um,” “uh,” “you know,” etc., then lets you delete them, ignore them, or remove them from the transcript only while leaving the audio intact. help.descript.comwww.capcut.comwww.timebolt.io
  • Good for: podcasts, interviews, YouTube videos
  • Can edit video by editing the transcript
  • Has an “Edit for Clarity” tool that goes beyond fillers and identifies wordiness, retakes, and unnecessary sections. www.descript.com

• CapCut — has an AI filler-word detector integrated with auto captions. It can identify things like “um,” “uh,” “like,” and “you know,” and you can remove them. www.capcut.com
  • Good if you're already editing short-form/social videos
  • Particularly convenient for caption + video editing together

• WordCut — a lightweight, browser-based option. It generates a timestamped transcript and lets you remove filler words, pauses, and sentences by editing the text. It claims to keep video processing in the browser. wordcut.video

• TimeBolt / UMCHECK — more specialized toward aggressively finding fillers and bad takes in video. It uses audio/waveform information alongside transcription and provides frame-accurate cuts and captions. www.timebolt.io

If your specific goal is subtitle suggestions

There's an important distinction:

Audio edit: “Remove ‘um’ from the speaker's recording.”

Subtitle edit: “Keep the speaker saying ‘um,’ but don't display ‘um’ in the subtitles.”

Descript can do the second one by removing the word from the transcript while leaving the audio unchanged. help.descript.com

So if you're building a workflow where an editor uploads a transcript/video and wants the tool to say:

Detected: “Um, I think we should, like, probably go tomorrow.”

and suggest:

Subtitle: “I think we should probably go tomorrow.”

while giving the editor Accept / Reject controls, that's very doable.

If you're asking because you want to build a tool like this, I can also outline exactly how I'd design the filler-word detection + subtitle-suggestion pipeline.

Yes. There are several tools that can detect filler words (like “um,” “uh,” “you know,” “like,” repeated phrases) in transcripts and either remove them or suggest cleaner subtitles/captions.

Some popular options:

For the specific workflow “upload transcript → detect filler words → suggest subtitle improvements → optionally remove them”, Descript is probably the closest match. It treats the transcript as the editing interface, highlights filler words, and lets you apply removals while keeping the video/audio synchronized.

If you are building your own tool, the typical pipeline is:

1. Speech-to-text (e.g., Whisper-based transcription)
2. Filler detection model/rules:
   • lexical fillers: “um,” “uh,” “like,” “you know”
   • repetitions: “I I think”
   • false starts: “I was— I mean…”
   • long pauses
3. Subtitle optimization:
   • shorten captions to reading speed limits
   • split lines naturally
   • preserve speaker intent
4. Human approval before applying edits

For a developer building this, the hard parts are usually not detecting “um” and “uh,” but deciding when words like “like” are filler versus meaningful (“I like pizza”).

Yes. There are several tools that do exactly this, depending on whether you want to remove fillers from the audio/video, remove them only from subtitles, or suggest cleaner wording.

Best options

1. Descript — probably the closest match

• Automatically detects fillers such as “um,” “uh,” “you know,” etc.
• Lets you delete, ignore, replace, or remove them from the transcript only.
• Because the transcript is linked to the video/audio, deleting text can automatically edit the media too.
• Its Edit for Clarity feature goes beyond fillers and can identify wordiness, tangents, retakes, and other unnecessary speech. help.descript.com
• Importantly, Descript supports removing a filler from the transcript without removing the corresponding audio/video, which is useful if you want cleaner subtitles while preserving natural speech. www.descript.com

2. CapCut — good for social/video captions

• Auto-generates captions.
• Detects filler words such as “um,” “uh,” “like,” and “you know.”
• Lets you review and delete detected fillers while working with the captions/video. www.capcut.com

3. WordCut — interesting free option

• Turns speech into a timestamped transcript.
• Lets you remove filler words, pauses, and entire sentences by editing the transcript.
• Runs much of the processing in the browser and exports the edited video. wordcut.video

4. UMCHECK / TimeBolt — more specialized

• Specifically focuses on detecting fillers and bad takes.
• Uses audio/waveform information alongside transcription, rather than relying solely on the transcript.
• Particularly interesting for long interviews, podcasts, and talking-head videos. www.timebolt.io

If you're looking for something slightly different

If by “subtitle suggestions” you mean:

Transcript: “Um, I think, like, we should probably, you know, launch it next week.”

and you want the tool to suggest:

“I think we should launch it next week.”

rather than simply deleting “um,” “like,” and “you know,” then Descript's Edit for Clarity is closer to what you want. It can clean up wordiness and awkward sections rather than merely applying a fixed filler-word list. help.descript.com

If you're building a tool/product yourself, there's also a very doable workflow where an AI analyzes the transcript and returns something like KEEP / REMOVE / REPLACE / REPHRASE for every questionable phrase, while preserving the original timestamps.

Yes. There are several tools that do exactly this, especially if you're working with video/podcast transcripts and subtitles.

Best options

1. Descript — probably the closest match

• Automatically detects filler words such as “um,” “uh,” “you know,” “like,” “I mean,” etc.
• You can review each detection and choose delete, ignore, replace with a gap, or remove from the transcript only. help.descript.com
• Importantly, it supports removing a filler word from the transcript without removing it from the underlying audio/video, which is useful when you want clean subtitles but natural speech. feedback.descript.com
• It also has Edit for Clarity, which goes beyond individual fillers and can identify wordiness, retakes, and unnecessary sections. help.descript.com

2. Adobe Premiere Pro

• Has text-based editing built around the transcript.
• Its Text-Based Editing includes filler-word detection and bulk removal workflows. community.adobe.com
• Better choice if you're already doing your video editing in Premiere.

If your goal is specifically clean subtitles without changing the spoken video

I'd lean toward Descript. Your workflow could be:

Video → transcription → detect fillers → remove fillers from transcript only → export clean captions/SRT

That avoids the common problem where deleting “um” from the audio creates a noticeable jump cut, while still giving viewers cleaner subtitles.

If you tell me what you're starting with (MP4, SRT, TXT transcript, Premiere project, etc.), I can also recommend the best tool/workflow for that specific setup.

Yes. There are several tools that do exactly this, especially if you're working with video/podcast transcripts.

Best fits

1. Descript — probably the closest match

• Automatically detects filler words like “um,” “uh,” “you know,” etc.
• Lets you delete the filler from both audio/video and transcript, or remove it from the transcript only while leaving audio intact. help.descript.com
• Its Edit for Clarity feature goes beyond individual fillers and can identify wordiness, tangents, retakes, and unnecessary chatter. www.descript.com
• It also has transcript-based video editing, so deleting text effectively edits the underlying media.

Descript's filler-word tool

2. Riverside

• Detects filler words automatically.
• You can choose Cut, Mute, or Smart removal, depending on whether you want the timeline shortened or preserved. www.descript.comsupport.riverside.comfeedback.descript.com
• Has captions/subtitles and transcript editing built into the same workflow. support.riverside.com
• Its Find Fluff feature can also identify unnecessary sections beyond simple “um/uh” fillers. support.riverside.com

If what you specifically mean is:

Upload transcript → automatically identify filler words → suggest cleaner subtitle wording → let me accept/reject each suggestion without necessarily changing the audio

That's a slightly more specific use case. Descript is close, but its native workflow is more focused on editing the underlying media. In fact, users have specifically requested a feature for removing fillers from captions while leaving the audio untouched. feedback.descript.com

If you're building a tool/product around this idea, though, I can also point you toward the APIs/models that would let you do filler detection + context-aware subtitle rewriting + optional removal automatically.