How the audio to text converter works
The audio to text converter reads your file in the browser, shows its length and price, and then sends the audio in parts of up to 2 minutes to the Whisper large-v3-turbo AI speech model on Cloudflare Workers AI. Each part comes back with timestamps, and the parts are joined into one transcript you can copy or download.
- Choose a recording
Pick an audio or video file (MP3, M4A, WAV, MP4, MOV, OGG, FLAC and more, up to 120 minutes). Your browser reads the sound and shows the length and the credit price before anything is uploaded.
- Set the language and names
Leave the language on Auto-detect or pick it. Optionally type names and technical terms that appear in the recording so Whisper spells them the way you do.
- Transcribe
Tick the consent box and press Transcribe. Long files go in 2-minute parts; keep the tab open until the last part is done.
- Copy or download
Switch between plain text and timestamps, copy the text, or download TXT, SRT or VTT. The transcript stays in your account.
Audio to text examples from public-domain recordings
The audio to text examples below are four public-domain recordings, each cut to under a minute: an outdoor speech, a 1941 radio broadcast, a two-person space radio call and a studio podcast. Next to each one is the transcript the open-source Whisper large-v3-turbo model produced when run offline, left unedited so you can see the mistakes. TranscribeBolt runs the same model on Cloudflare Workers AI, with different settings, so its text for the same clip can differ.
1962 Moon speech at Rice University (outdoor stadium, applause)
A 60-second stretch of an outdoor speech with crowd noise and applause. The famous lines came through, but "We choose to go to the moon" appears one time more than it was said, and "postpone" came out as "postponed".
may never come again. But why some say the moon? Why choose this as our goal? And they may well ask, why climb the highest mountain? Why 35 years ago fly the Atlantic? Why does Rice play Texas? We choose to go to the moon. We choose to go to the moon. We choose to go to the moon in this decade and do the other things, not because they are easy, but because they are hard. Because that goal will serve to organize and measure the best of our energies and skills. Because that challenge is one that we're willing to accept, one we are unwilling to postponed and one we intend to win and the others too.
Apollo 13 air-to-ground radio (two speakers, static)
A 40-second radio exchange between the crew and Mission Control. Each radio call became its own short line, but nothing says who is speaking, and "main B bus undervolt" came out as "main B plus undervolt".
Okay, Houston, we've had a problem here. This is Houston, say again, please. Houston, we've had a problem. We've had a main B plus undervolt. Roger, main B undervolt. Okay, stand by 13, we're looking at it. Okay, right now, Houston, the voltage is looking good. And we had a pretty large bang associated with the caution and warning there. And as I recall, main D was the one that had an amp spike on it once before.
Space-station podcast interview (two voices, studio sound)
A 56-second studio interview, host then guest. The show name "Houston We Have a Podcast" was split into a sentence, long answers became long lines, and a one-word line at 0:00 is left over from where the clip was cut.
Yeah. Tracy, welcome to Houston. We have a podcast. We are so happy to have you here. Thank you. It's my pleasure. So before we get into the topic of the space station, I want our listeners to get to know you. So tell us a little bit about you, your background, and maybe when you joined NASA, what things were like at that time. What was human spaceflight exploration looking like at that time? I should go back to the year of 1997. I had finished up my PhD and submitted my dissertation in the month of June. And I think it was the next month, July, when I submitted my application for the astronaut program. And then I went on and did a postdoctoral fellowship in Southern California at UC Irvine, and it was during that period of time that I got the notice that I was being offered an interview. And so that kicked it off.
1941 address to Congress (old radio recording)
A 57-second clip of a scratchy 1941 broadcast with slow, formal delivery. Long pauses split the speech into short timestamped lines; the opening dropped the "and" in "and of the House of Representatives".
Mr. Vice President, Mr. Speaker, members of the Senate, of the House of Representatives, yesterday, December 7, 1941, a date which will live in infamy, the United States of America was suddenly and deliberately attacked by naval and air forces of the Empire of Japan. The United States was at peace with that nation and at the solicitation of Japan was still in conversation with its government and its emperor looking toward the maintenance of peace in the Pacific.
Can you convert audio to text for free?
No, audio to text on TranscribeBolt is not free: new accounts get no free credits right now, so a credit pack is needed before the first transcript. After that every file costs 2 credits per started minute with a 5-credit minimum, so 10 minutes is 20 credits, about $0.20. The smallest pack is $9.90 for 900 credits (about 450 minutes). Choosing a file and seeing its price costs nothing, and the text has no watermark.
If you only need a phone voice message or a short memo, the free built-in transcripts in WhatsApp and the iPhone Voice Memos app may be enough; the step-by-step guide covers them. How to transcribe audio to text on iPhone and WhatsApp
Can you use audio to text without signing up?
No. TranscribeBolt's audio to text converter needs a free account before it transcribes, because each file is paid with credits. You can open the page, choose a file and see its length and price without an account. The account is free to create, and audio to text works online in the browser, with no app or extension to install.
Does audio to text work with iPhone voice memos (M4A)?
Yes. Audio to text on TranscribeBolt accepts the M4A files the iPhone Voice Memos app records. In Voice Memos, tap the recording, then share it and choose Save to Files; then open this page in Safari, tap "Choose an audio or video file" and pick it from Files. A 40-minute lecture is 80 credits, about $0.80.
The page works in mobile Safari and Chrome on Android. Very long recordings take a lot of memory to read on older phones; if a 120-minute file will not open, try it on a computer.
Audio to text pricing and credits
Audio to text costs 2 credits per started minute of audio on TranscribeBolt, with a 5-credit minimum per file: 20 credits for 10 minutes, 120 credits for an hour. At the Pro pack rate a credit is $0.01, so an hour is $1.20. From a zero balance the smallest purchase is the $9.90 Starter pack (900 credits, about 450 minutes), a one-time payment with no subscription and no watermark.
| Audio length | Credits | Price | Watermark |
|---|---|---|---|
| Up to 1 minute | 5 | $0.05 | None (text only) |
| 3 minutes | 6 | $0.06 | None (text only) |
| 10 minutes | 20 | $0.20 | None (text only) |
| 30 minutes | 60 | $0.60 | None (text only) |
| 60 minutes | 120 | $1.20 | None (text only) |
| 120 minutes | 240 | $2.40 | None (text only) |
Price at $0.01 per credit (Pro pack). Smaller packs cost slightly more per credit.
| Pack | Price | Credits | Covers |
|---|---|---|---|
| Starter | $9.90 | 900 | ≈ 450 minutes of audio (7.5 hours) |
| Plus | $24.90 | 2,400 | ≈ 1,200 minutes of audio (20 hours) |
| Pro | $49.90 | 4,990 | ≈ 2,495 minutes of audio (41.6 hours) |
What audio to text can't do well
The audio to text converter does not label speakers: in a meeting or interview you get the words in order but not who said them. It is not live captioning, it does not take YouTube or social links, and it does not translate. Heavy background noise, people talking over each other and strong accents cause wrong words, and Whisper can repeat a line or invent text during long silence or music.
- Speaker names or "Speaker 1 / Speaker 2" labels: not available.
- Longest file: 120 minutes and 500 MB. Split longer recordings first.
- Links (YouTube, TikTok, Google Drive): not supported, upload the file itself.
- Translation: not available; the transcript is in the spoken language.
- Accuracy: no promise of a word error rate; check names, numbers and quotes before you publish them.
Audio to text FAQ
Which languages can it transcribe?
Whisper large-v3-turbo is a multilingual model; OpenAI released Whisper for roughly 100 languages. The language menu lists 22 common ones (including Spanish, Portuguese, Hindi, Bengali, Urdu, Arabic, Japanese, Korean and Chinese) plus Auto-detect, which lets the model pick. A recording that mixes languages may come back written in one of them, so pick the main language when you know it.
Is my recording stored?
TranscribeBolt does not save your audio. Your browser sends it in parts to our server, which passes each part to the speech service and drops it after the reply. The transcript text is saved to your account so you can come back to it. The speech provider handles audio under its own terms, so do not upload recordings you are not allowed to share with a third party.
Can I transcribe the audio of a video file?
Yes. MP4, MOV and WebM videos are accepted; your browser takes the sound track out and only the audio is transcribed. The video itself is never uploaded, and the 500 MB limit is checked on the file you choose.
What happens if a transcript fails?
If the speech service cannot transcribe a part of your file, the job stops and all of its credits are refunded automatically. If you close the tab before any audio arrives, the job is refunded within 30 minutes; if some parts arrived, you keep the transcript of those parts.
Can I use the transcript commercially?
Yes, the text is yours to use, for show notes, articles, captions or research. You are responsible for having the right to record and transcribe the audio, and for checking quotes before you publish them.