# audiohighlight

> AI transcription that arrives finished — speaker labels you fix in bulk, every word linked to its second of audio. $0.25 per minute, pay per file, no subscription. private mode runs locally in your browser; nothing uploaded.

This site is the pre-launch fake-door for a transcription product positioned for two audiences: (1) journalists and reporters working interviews, video b-roll, and press conferences, and (2) podcasters, bloggers, and authors turning recordings into copy. Buyers who need a transcript fast and refuse to upload sensitive or unpublished audio to a third-party server.

## What we do differently

- **Cleanup time is the metric.** Word error rate doesn't predict whether a transcript saves you time. We benchmark cleanup-time-per-audio-hour on a published corpus instead.
- **Private mode runs in your browser.** WebGPU + whisper, no upload, no third-party server. Audio never leaves the device.
- **Format-first exports.** SRT/VTT captions for video and YouTube, show-notes drafts for podcasters, blog-post drafts for writers, publication-ready quotes for reporters with a deadline.
- **Pay per file, $0.25 per minute, all features.** Same price for cloud and on-device. No subscription.

## Pages

### Core
- [Home](https://audiohighlight.com/): hero, transcript preview, use-case grid for our two ICPs.
- [Manifesto](https://audiohighlight.com/manifesto): the extended Edison thesis. Why this exists.
- [About](https://audiohighlight.com/about): why we built this, what we explicitly aren't.
- [Pricing](https://audiohighlight.com/pricing): $0.25/minute, pay-per-file, all features included, private mode same price as cloud.
- [Benchmark](https://audiohighlight.com/benchmark): the cleanup-time-per-audio-hour metric, methodology, audio corpus.
- [FAQ](https://audiohighlight.com/faq): the actual fifteen questions buyers ask, with the actual answers.
- [Transcript library](https://audiohighlight.com/transcript-library): sample finished transcripts across the audio our two ICPs actually record.

### Writing (long-form)
- [Blog index](https://audiohighlight.com/blog)
- [The cleanup tax](https://audiohighlight.com/blog/the-cleanup-tax): the hidden cost of every transcription tool.
- [WER is a useless buyer metric](https://audiohighlight.com/blog/wer-is-a-useless-buyer-metric): why word error rate doesn't predict your time.
- [How to audit a browser-based transcription tool](https://audiohighlight.com/blog/audit-a-browser-transcription-tool): five-minute network-tab audit.
- [The diarization fragility test](https://audiohighlight.com/blog/the-diarization-fragility-test): adversarial speaker-labeling tests across the major commercial tools.
- [Open-source audio tools we love](https://audiohighlight.com/blog/open-source-audio-tools-we-love): the audio software stack we actually use — audacity, ffmpeg, whisper, openwhispr, and a few less-obvious favorites.

### Buyer-keyword landings (how-to pages)
For journalists and reporters:
- [Transcribe an interview](https://audiohighlight.com/transcribe-interview)
- [Transcribe video content](https://audiohighlight.com/transcribe-video-content)
- [Transcribe a press conference](https://audiohighlight.com/transcribe-press-conference)

For podcasters, bloggers, and authors:
- [Transcribe a podcast](https://audiohighlight.com/transcribe-podcast)
- [Transcribe dictation](https://audiohighlight.com/transcribe-dictation)
- [Transcribe a meeting](https://audiohighlight.com/transcribe-meeting)
- [Transcribe a Zoom recording](https://audiohighlight.com/transcribe-zoom-recording)

### Free in-browser tools (the chain we give away)
The chain is **ffmpeg → LLM → time-aligned transcript → remotion**. Each link is a free tool that runs locally in the browser — no upload, no account, no third-party server.
- [Tools index](https://audiohighlight.com/tools): the full chain, organized by tier.
- [Voice recorder](https://audiohighlight.com/tools/voice-recorder): record from your mic in the browser. Live.
- [Extract audio from video](https://audiohighlight.com/tools/extract-audio-from-video): tier-1 ffmpeg.wasm tool. Drop a video file, get an mp3 back. Live.
- [Trim audio](https://audiohighlight.com/tools/trim-audio): tier-1 ffmpeg.wasm tool. Cut a segment out of an audio file by start/end timestamps. Live.
- [Convert audio formats](https://audiohighlight.com/tools/convert-audio): tier-1 ffmpeg.wasm tool. Convert between mp3/wav/m4a/ogg/opus/flac. Live.
- [Compress audio](https://audiohighlight.com/tools/compress-audio): tier-1 ffmpeg.wasm tool. Shrink a file to a chosen mp3 bitrate. Live.
- [Transcribe without uploading](https://audiohighlight.com/tools/transcribe-without-uploading): tier-3 chain piece. WebGPU + Whisper running locally. Preview at launch.

### Private transcription (on-device, no upload)
- [Private transcription overview](https://audiohighlight.com/private-transcription)
- [For journalists](https://audiohighlight.com/private-transcription/journalism): source protection without a subpoena-able vendor.
- [For podcasters](https://audiohighlight.com/private-transcription/podcasting): embargoed announcements, off-the-record passages, privacy-conscious guests.
- [For bloggers](https://audiohighlight.com/private-transcription/bloggers): pre-publication drafts that shouldn't sit on a third-party server.
- [For authors](https://audiohighlight.com/private-transcription/authors): unpublished manuscript material and confidential research interviews.
- [Transcribe without uploading](https://audiohighlight.com/tools/transcribe-without-uploading): the browser tool itself.

### ICP-relevant export formats
- [Formats overview](https://audiohighlight.com/formats)
- [SRT / VTT captions](https://audiohighlight.com/formats/captions): captions for video, accessibility, and YouTube.
- [Show notes export](https://audiohighlight.com/formats/show-notes): podcast-ready show notes from a finished transcript.
- [Blog post draft export](https://audiohighlight.com/formats/blog-post): markdown + .docx draft from interview or dictation audio.
- [Publication-ready quotes](https://audiohighlight.com/formats/clean-quotes): light-touch cleanup, timestamps preserved for fact-check.

### Comparisons
- [vs index](https://audiohighlight.com/vs): comparison hub.
- [vs temi](https://audiohighlight.com/vs/temi): the budget option whose accuracy regressed.
- [vs rev](https://audiohighlight.com/vs/rev): rev's two products and where they stop fitting.
- [vs sonix](https://audiohighlight.com/vs/sonix): the workspace-first alternative without the subscription cliff.
- [vs otter](https://audiohighlight.com/vs/otter): otter is a meeting bot. we aren't. different categories.
- [vs descript](https://audiohighlight.com/vs/descript): descript edits the content. we finish the transcript.

### Other audience pages (URLs alive but not surfaced from homepage nav)
The transcription product fits other audiences too — these pages remain available for search-driven discovery, but the homepage nav is focused on our two primary ICPs.
- [For therapists](https://audiohighlight.com/private-transcription/therapy): HIPAA-aware, BAA-free.
- [For physicians and medical scribes](https://audiohighlight.com/private-transcription/medical): clinical dictation and notes, BAA-free.
- [For lawyers](https://audiohighlight.com/private-transcription/legal): privileged audio without a vendor in the chain.
- [For UX and product researchers](https://audiohighlight.com/private-transcription/ux-research): NDA-bound user interviews, pre-launch product feedback.
- [Transcribe a deposition](https://audiohighlight.com/transcribe-deposition): legal-cluster how-to.
- [Jefferson notation](https://audiohighlight.com/formats/jefferson-notation): for conversation analysts.
- [Deposition format](https://audiohighlight.com/formats/deposition-formatter): 25-line paginated, court-ready.
- [NVivo export](https://audiohighlight.com/formats/nvivo-export): for qualitative researchers using NVivo.
- [ATLAS.ti export](https://audiohighlight.com/formats/atlas-ti): same engine, ATLAS.ti's import schema.
- [MAXQDA export](https://audiohighlight.com/formats/maxqda): same engine, MAXQDA's import schema.

### Site
- [Privacy](https://audiohighlight.com/privacy)
- [Terms](https://audiohighlight.com/terms)
- [Atom feed](https://audiohighlight.com/feed.xml)
- [Sitemap](https://audiohighlight.com/sitemap.xml)

## What we are not

- Not a meeting bot. No calendar OAuth, no auto-join. File uploads only.
- Not enterprise software. No SSO, no SCIM, no procurement cycles at launch.
- Not a transcription marketplace. We don't recruit human transcribers.
- Not a court reporter substitute. Court reporter transcripts remain the official record.
- Not a podcast/video editing studio. Descript is. We finish the transcript.

## Status

Pre-launch. The waitlist captures intent; the working build ships when it can run a real file end-to-end. When in doubt about whether something on this site is shipped or planned: assume planned.