Words & language · Free
Speech to text – free online dictation
Dictate emails, notes, reports or ideas and get clean text – free speech to text in your browser. Your voice is processed on your device, never uploaded, with no time limit.
- No account
- Runs on your device
- Phone & desktop
Speak instead of typing
Stays on your device
Drop an audio or video file here
MP3, M4A, WAV, OGG, OPUS, FLAC, MP4, MOV, WEBM and more · long files welcome
Recording…
0:00Writing your transcript
·
- Reading the file
- Loading the speech model
- Transcribing
Text so far
Your transcript
Click any sentence to correct it.
Private by design. The recording is processed on your device and never uploaded. Only the speech model is downloaded, once, from Hugging Face.
Fastest on a computer. With a recent graphics chip (WebGPU) an hour of audio takes a few minutes. Phones work too, a little slower.
How to dictate with Transcript Studio
Press Record
Allow the microphone once. A level meter shows that your voice comes through.
Speak naturally
Talk in whole sentences. You don’t need to say “comma” or “full stop”: Whisper adds most punctuation by itself.
Stop and use the text
Press Stop and transcribe. Correct anything, then copy the text or save it as Word.
What people use Transcript Studio for
- Interviews and research – an hour of interview takes four to six hours to type; here it takes minutes.
- Lectures and lessons – turn recordings into notes you can search and quote.
- Meetings and calls – a written record instead of rewinding.
- Videos and social media – subtitles for YouTube, TikTok and Instagram, where most people watch without sound.
- Dictation – speak emails, reports or ideas instead of typing them.
- Accessibility – make audio readable for people who are deaf or hard of hearing.
Why no upload matters
Most online transcription services upload your recording to their servers and limit free use to a few minutes. Transcript Studio runs OpenAI’s open Whisper model inside your browser instead. The audio never leaves your device, which matters for confidential interviews, medical or legal dictation and company meetings – and there is no minute limit.
Subtitles in SRT and VTT
Every sentence comes with its time. The SRT and VTT exports split it into readable subtitle lines (you choose the line length) that YouTube, Vimeo, Premiere and DaVinci Resolve accept. On a video you can preview the subtitles right on the picture before you save them.
How accurate is it?
Whisper understands about 99 languages and is very good with clear speech. Accuracy drops with music, strong background noise or several people talking at once. Choose Accurate for difficult recordings, set the language instead of automatic detection, and proofread names and numbers before you publish.
Languages and translation
English, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Turkish, Arabic, Chinese, Japanese and many more are recognised. Tick Translate into English to get an English transcript of a recording in another language in one step.
More Transcript Studio pages
Frequently asked questions
Is it better than the dictation built into my phone?
The dictation built into iPhone or Android is great for short messages. Transcript Studio suits longer dictation: you can play the recording back, see every sentence with its time and export to Word or text, privately on your phone or computer.
Does it work offline?
After the speech model has been downloaded once, transcription runs without a server. The page itself needs to load from the internet.
Is Transcript Studio really free?
Yes. There is no account, no subscription, no watermark and no limit on minutes or files. It is funded by the site, not by your recordings.
Are my recordings uploaded?
No. Decoding and speech recognition run in your browser. Only the speech model is downloaded once from Hugging Face; your audio is never sent anywhere.
Which file formats work?
Anything your browser can play: MP3, M4A, AAC, WAV, OGG, OPUS, FLAC, WEBM and video files such as MP4 and MOV. Files can be up to 1.5 GB on a computer and 400 MB on a phone.
How long does it take?
On a laptop with WebGPU an hour of audio usually takes a few minutes with Fast. Without WebGPU and on phones it takes longer; you see the text appear and a time estimate while it works.
Can I make subtitles for YouTube or TikTok?
For YouTube, yes: export SRT or VTT and upload the file with your video, or import it into your editing program. You can set how many characters fit on one line. TikTok and Instagram only show subtitles reliably when they are part of the picture, so burn them in with Subtitle Studio.
Can it recognise different speakers?
Not yet. The transcript is split into timed sentences without speaker names; add names while you correct the text.
Take a break
