Open-source and self-hosted

Private transcription for long conversations.

Macaw Transcribe helps you turn interviews, podcasts, research calls, and spoken notes into transcripts organized by speaker, you can review while processing continues.

First release optimized for macOS on Apple Silicon
Macaw Transcribe
1h 23m2 speakers detected

interview-recording.mp3

47.3 MB1h 23m

Transcribing locally0%

Speaker Timeline

speakers detected

Speaker A
Speaker B

Transcript Review

2 completedTranscribing locally
Speaker ACompleted

So when we started looking at the data, it became clear that the patterns we'd been seeing weren't just noise. There was something systematic happening.

Speaker BCompleted

That's fascinating. Can you walk me through what you found? I'm particularly interested in how you identified those patterns initially.

Speaker AReviewing
Speaker BTranscribing locally

Long conversations hold valuable knowledge

But most transcription tools are either too opaque, too SaaS-dependent, or difficult to review carefully.

Long recordings are hard to review

Hours of audio turn into walls of text. Scrubbing through recordings to verify transcripts is tedious and error-prone.

Speaker context gets lost

Without clear diarization, conversations become confusing. Who said what matters, especially in interviews and research.

Private audio deserves private infrastructure

Sensitive conversations shouldn't require uploading to third-party servers. Your audio should stay under your control.

Local hardware is often underused

Many self-hosted transcription workflows are not tuned for modern Mac hardware, turning powerful machines into slow, inefficient transcription boxes.

Errors show up too late

You often discover accuracy problems only after the full transcript is done. Fixing one bad section should not mean starting the whole recording over.

A transcription workflow built for control, review, and accuracy

Macaw Transcribe is designed around how you actually work with long-form audio.

Self-hosted by design

Run on your own machine or private server. Your audio never leaves your infrastructure.

Progressive transcription

Long recordings are processed in chunks. Start reviewing before everything is finished.

Speaker-aware review

Built-in diarization organizes conversations by speaker. Review context, not just text.

Chunk-level review and retries

Fix problems where they happen. Retry difficult sections without restarting the whole recording.

Full transcript assembly

Chunks are joined into a coherent document. Overlaps handled, context preserved.

Export-ready output

Clean, formatted transcripts ready for your workflow. Export when you need them.

From audio to clean transcript

A practical workflow designed for long recordings and careful review.

1

Upload audio

Drop your recording into the app. Supports common audio formats.

2

Normalize recording

Audio is cleaned and prepared for optimal transcription quality.

3

Detect speakers

Speaker detection identifies who speaks when throughout the recording.

4

Split into chunks

Long audio is split into overlapping chunks so each section can be processed and reviewed safely.

5

Transcribe progressively

Chunks are transcribed progressively, so useful review can begin before the full recording is finished.

6

Review, edit, and export

Verify transcripts, adjust speaker labels, and export your final document.

Built for people who work with long-form audio

Whether you're conducting research, producing content, or just want control over your transcriptions.

Researchers

Reviewing interviews

Process hours of qualitative research interviews. Review transcripts chunk by chunk, with speaker labels intact for analysis.

Journalists

Working with sources

Transcribe confidential conversations without third-party access. Verify quotes against the original audio.

Podcasters

Editing long episodes

Turn multi-hour recordings into searchable transcripts. Edit show notes while the rest of the episode processes.

Writers

Spoken notes to drafts

Capture ideas through voice recordings and convert them to editable text. Review and refine your spoken thoughts.

Therapists & coaches

Reviewing private sessions

Turn long conversations into reviewable notes while keeping sensitive audio under your own control.

Privacy-focused teams

Avoiding cloud transcription

Keep sensitive meeting recordings on your own infrastructure. No data leaves your control.

Not just transcription. A review workspace.

Most transcription tools focus on output. Macaw Transcribe focuses on the entire workflow.

Feature
Macaw
Output-first tools
Designed for long-form audio
Built around review, not just output
Speaker-aware from the start
Self-hosted and transparent
Progressive chunk-based workflow
Review while processing
Chunk-level correction and retries

The key difference: Macaw Transcribe is designed for the work that happens after transcription starts: reviewing, correcting, retrying difficult sections, and exporting a clean final transcript.

Open Source

Built on transparency and control

Macaw Transcribe is being built as an open-source, self-hosted foundation for people who want control over their transcription workflow. No vendor lock-in. No black boxes.

Run it locally

Install on your laptop, workstation, or private server. No cloud transcription service required.

Inspect the code

Every line is visible. Understand exactly how your audio is processed.

Own your workflow

Customize, extend, or integrate with your existing tools. It's your infrastructure.

Contribute to the roadmap

Shape the product through issues, pull requests, and community feedback.

terminal
$ git clone https://github.com/samuelkb/MacawTranscribe.git
Cloning into 'MacawTranscribe'...
✓ Repository cloned successfully
$ cd macaw-transcribe && docker compose up
$ python manage.py run_worker_supervisor && python manage.py run_workspace_supervisor
✓ Worker supervisor started
✓ Workspace supervisor started
$ gunicorn MacawTranscribe.wsgi.application
Starting services...
✓ Macaw Transcribe running on localhost:8000

Local setup is being simplified

In active development

Be part of building it

Macaw Transcribe is currently being built. Join the waitlist to follow progress, test early releases, and help shape the product.

No spam. Occasional updates on development progress and early access.

Or follow development directly

Follow on GitHub

Progress updates

Regular development logs

Early access

Test before public release

Direct input

Shape features and priorities