Drop your MP3 or M4A here (files over 500 MB continue on the upload page). You get a text transcript with word-level timestamps in about two minutes per hour of audio, with a summary and DOCX, SRT, VTT, TXT and JSON exports. Free preview with no signup; packs from $4 for 300 minutes, and credits never expire.
How it works
Upload in the browser: no local install, no Python, no CLI.
64 kbps voice memo or 320 kbps studio file — both work. Up to 10 hours or 5 GB per file; files over 500 MB continue on the upload page. Other formats have their own notes: WAV, M4A, MP4.
A 30-minute MP3 transcribes in about a minute, a three-hour one in about six. The opening minutes are a free preview; word-level timestamps are included.
DOCX for a writer handoff, SRT/VTT for captions, TXT for a quick paste, JSON for downstream tooling. All included on every pack.
What you can do with the transcript
Every MP3 transcript opens with an AI sidebar — search, summarize, chat, export.
Click a line to jump to the second in the audio player. Search a name or phrase — every match returns with a timestamp.
One click produces a chapter-marked summary, key quotes, and a short social blurb.
Ask any question of the audio — “what did the speaker conclude?” or “list every action item.” The AI answers with quoted timestamps.
DOCX, SRT, VTT, TXT, JSON — every format included on every pack, no Pro paywall.
MP3 numbers
MP3 is the second most common file people bring us: 642 of the 1,934 file uploads finished in the last 90 days, a third of them. In the last 30 days 482 MP3s were transcribed. The typical one ran 13.6 minutes and weighed 9 MB, and half of them came back within 37 seconds of landing.
| MP3s, last 30 days | Measured |
|---|---|
| Files transcribed | 482 |
| Median length | 13.6 minutes |
| Nine in ten were shorter than | 79 minutes (90-day window) |
| Median file size | 9 MB, about 58 MB per audio hour |
| Median time from upload to transcript | 37 seconds |
From our production record, measured 2026-10-27 (30 days) and 2026-10-26 (90 days, internal and test accounts excluded). Method: what people bring and turnaround.
MP3 size is set by the bitrate, not the content: kilobits per second times 3,600 seconds. Every row fits easily under the 5 GB limit; the 10-hour limit is the one that applies.
| Bitrate | Typical source | One hour | Ten hours |
|---|---|---|---|
| 64 kbps | Voice memos, dictation, phone recorders | about 29 MB | about 288 MB |
| 128 kbps | Podcasts, most recorder exports | about 58 MB | about 576 MB |
| 192 kbps | Higher-quality podcast exports | about 86 MB | about 864 MB |
| 320 kbps | Music-grade exports | about 144 MB | about 1.4 GB |
The box at the top takes files up to 500 MB. A bigger MP3, such as ten hours at 192 kbps, continues on the main upload page.
Packs are counted by the minute of audio, one-time, with no subscription, and the minutes never expire.
| Recording | Minutes used | $4 pack (300 min) | $18 pack (3,000 min) | $32 pack (6,000 min) |
|---|---|---|---|---|
| Typical MP3, 14 minutes | 14 | $0.19 | $0.08 | $0.07 |
| One-hour interview | 60 | $0.80 | $0.36 | $0.32 |
| Three-hour lecture | 180 | $2.40 | $1.08 | $0.96 |
The $4 pack covers about 21 typical MP3s, or five one-hour files. For a team that transcribes every week, Workspace is $19 a month for up to 300 hours. See pricing.
The free preview covers the opening minutes, not the whole file. Speakers are named in the summary, key quotes and chapters, not on every line of the transcript or in the SRT. A file over 10 hours needs splitting first. Heavy music under speech, or a 32 kbps phone copy of a noisy room, will cost accuracy no matter which service transcribes it.
Pricing
Every MP3 gets a free preview, and in 31 countries, or for an hour or more in most others, a summary you can read before paying. A pack opens the full transcript and every export: one-time purchases, counted by the minute of audio, with no monthly subscription.
Long recordings: files up to 10 hours or 5 GB, including stream audio downloaded after the broadcast ends. A 3-hour recording uses 180 minutes, about $1.08 of the $18 pack or $2.40 of the $4 pack. The language is detected automatically.
Pick a pack to open secure checkout. Full pricing
Where is the MP3 from?
An MP3 off a voice recorder, an episode at a web address and an export from an editor each have their own shortcut. For a short clip, the MP3 to text converter is the quick route; this page is for long recordings.
3,332 recordings, 2,490 audio hours, 65 languages detected. An hour of audio came back in 3.1 minutes for half of the recordings and within 17.5 minutes for nine in ten, measured from the moment the file landed to the finished transcript. Refreshed 2026-10-28; method on how we test.
FAQ
Every MP3 starts with a free preview of its opening minutes, with no signup; in 31 countries, or for recordings of an hour or more in most others, the summary is free to read too. The full transcript and the exports need a pack: $4 for 300 minutes, $18 for 3,000 or $32 for 6,000, and the minutes never expire.
Up to 10 hours or 5 GB per file. At 128 kbps an hour of MP3 is about 58 MB, so the 10-hour limit is reached long before the 5 GB one. Longer recordings can be split and uploaded as two files.
60 minutes of credit: $0.80 from the $4 pack, $0.36 from the $18 pack or $0.32 from the $32 pack. The median MP3 people sent us in the last 30 days ran 13.6 minutes, about 18 cents from the $4 pack.
99 languages, detected automatically, so you do not pick one before uploading. A transcript can also be translated into any of the 99 languages with the timestamps kept.
Yes. SRT and VTT caption files come with every pack alongside DOCX, TXT and JSON, built from the same timestamps as the transcript.
Drop the MP3 onto the upload box above. A text transcript with word-level timestamps returns in about a minute per 30 minutes of audio. Preview with no signup; a pack from $4 opens the full transcript.
No. 64 kbps voice memos through 320 kbps studio files all work. Lower bitrates with heavy background noise may dip in accuracy, but the typical range is handled well.
DOCX, SRT, VTT, TXT, and JSON — all included on every pack. No Pro-tier paywall on any export.
A 30-minute MP3 in about a minute, a 1-hour file in about two minutes.
In the summary, key quotes and chapters, yes: speakers are identified and named there. The transcript itself is a list of timestamped lines, so you can check any attribution in the summary against the audio.
Up to 10 hours or 5 GB per file. The box on this page takes files up to 500 MB; larger files continue on the upload page.
Yes. Your audio and transcripts are never used to train AI models.
Instant preview, no signup, no card. A free account keeps your transcripts in one library.
Upload a file or paste a link: see how our transcription services work and an audio transcription service that takes links or files.
This page in: Deutsch · Français · Español · Português · 한국어