Turbo mode
Turbo splits a long recording and sends the pieces to several speech engines at once, so an hour of audio comes back in a fraction of the time. A slice that fails is retried on another engine.
Alchemys App
Transcribe AI
Drop an audio or video file, paste a YouTube, Instagram or Facebook Reel link, or record from your microphone. Transcribe AI writes the transcript, labels each speaker, and gives you a summary, meeting minutes, translations and a chat about the recording — all editable and timestamped.
Sign in with Google to use it — new accounts get free starter credits, shared with every other Alchemys app.
What's newUpdated October 2026
Recently added to Transcribe
Faster on long files, and easier to find things in afterwards.
Collections
File recordings into named collections — a lecture series, a podcast, a project — and keep them together. Deleting a collection never deletes a transcript.
Search what was said
Search now looks inside every transcript, so a term spoken halfway through a lecture is found without opening it. Edits you make to a transcript are searchable too.
Continue right where you were
The Continue card on the home dashboard now reopens the exact transcript you last worked on, not just the app.
What you can do
Turn any recording into a working transcript
Seven transcription engines plus Auto and Turbo modes, one drop zone — Auto is the default, so drop a file in and let it run, or pick the engine that matches your file.
Capture it any way you like
Drop an audio or video file, paste a YouTube, Instagram or Facebook Reel link, or tap Record to capture straight from your microphone.
Pick the right engine for the job
Choose Auto (the default: our own GPU box first, a paid engine only when it is offline or slow), MiniMax ASR, Fun-ASR, Cloudflare Whisper, Chirp 3, SeedASR, Gemini or the self-hosted Alchemys Local engine — each has its own size or duration limit.
Identify who said what
Turn on speaker identification, or let Chirp 3 and SeedASR label speakers automatically as part of the transcription.
Get a summary and minutes for free
Every transcript comes with a summary and meeting minutes with decisions and action items you can send to Calendar or Notebook.
Translate the transcript
Translate the finished transcript into Simplified or Traditional Chinese, English, Japanese, Korean, French or German, right next to the original.
Export or keep chatting
Export to TXT, Markdown, SRT, VTT, JSON or Word, or open the Chat tab to ask questions about the recording.
Getting started
From recording to transcript in five steps
Everything below happens on the Transcribe AI home screen.
-
Add your audio
Drop an audio or video file onto the upload area, tap YouTube, Reels or URL to paste a link, or tap Record to capture live.
-
Choose an engine and language hint
Open the Engine panel and pick the engine that matches your file — Auto (the default) for most recordings and Cantonese, Fun-ASR for very large files, Chirp 3 or SeedASR for long ones, Gemini for Cantonese meetings — then set a language hint if you know it.
-
Turn on speaker identification if you need it
Switch on Identify speakers before you start — skip this for Chirp 3 and SeedASR, which label speakers automatically.
-
Review the result
Once it's done, use the Transcript, Summary, Minutes, Translation and Chat tabs to check names, fix a segment, or ask a question about the recording.
-
Export or send it on
Use Export to download TXT, Markdown, SRT, VTT, JSON or Word, or tap Send to Notebook or Add to Calendar straight from the minutes.
Good to know
Formats, limits and what stays where
The details that decide which engine you should pick.
- Supported files Audio and video — MP3, WAV, M4A, MP4, MOV, MKV, AVI and more, auto-converted for your engine
- Engine limits Auto (default, local GPU first) · MiniMax ASR (auto-sliced) · Fun-ASR ≤ 2 GB · Cloudflare Whisper ≤ 90 min · Chirp 3 ≤ 8 hr · SeedASR ≤ 5 hr · Alchemys Local ≤ 8 hr
- Export formats TXT, Markdown, SRT, VTT, JSON, Word (.docx), or copy to clipboard
- History Transcriptions stay in your history until you delete them
- Account Google sign-in required — shares your Alchemys credit balance
The Alchemys Local engine runs on a self-hosted GPU box, so it's only available while that machine is online — pick a hosted engine if it's offline.
FAQ
Questions people ask before their first transcript
Which languages does it support?
Speech-to-text and speaker identification are tuned for Cantonese, English and Mandarin, with language hints for Japanese and Korean too. Translation covers English, Simplified Chinese, Traditional Chinese (Hong Kong or Taiwan), Japanese, Korean, French and German.
Does it cost credits?
Yes. New accounts get free starter credits, then usage is billed per model from the same balance as every other Alchemys app — see the pricing page for current rates.
Can I transcribe a YouTube or Instagram link?
Yes — tap YouTube, Reels or URL and paste a public YouTube, Instagram or Facebook Reel link, or a direct link to an audio or video file.
What if my recording is too long for an engine?
Switch engines in the Engine panel. Fun-ASR (up to 2 GB), Chirp 3 (up to 8 hours) and SeedASR (up to 5 hours) all cover longer recordings than Cloudflare Whisper's 90-minute limit, and Auto (the default) and MiniMax ASR slice long audio automatically.
Do I need to sign in?
Yes. Google sign-in is required, and every transcript, summary and chat is kept in your history until you delete it.
Ready to transcribe your first recording?
Sign in with Google and drop a file — your first transcript is a click away.