Keep this tab open — you'll be redirected to your transcript.
PodcastsVoice memosInterviewsPhone recordings
Why MP3 is the easy case
MP3 is what most spoken audio already lives in: podcast downloads, voice recorder apps, old archives, exported calls. The model was trained on enormous amounts of compressed speech, so the MP3 codec itself costs essentially nothing in accuracy — a clean 64 kbps mono voice recording transcribes as well as a studio WAV.
Upload the file as-is. There is no need to convert, re-encode, or "improve" an MP3 before transcribing; re-encoding can only lose information.
From MP3 to notes, quotes, and subtitles
A 30-minute podcast episode fits the anonymous free tier; a full-length episode or a 2-hour interview needs a free account (1 hour) or Pro (10 hours). After processing, use the editor to fix names and jargon, then export TXT for notes and quotes or SRT/VTT if the audio will be published with video.
Frequently asked questions
Does a low MP3 bitrate hurt accuracy?
Barely, for speech. Anything from 64 kbps mono upward transcribes essentially as well as lossless audio — phone voice memos and podcast downloads are well above that. Below ~32 kbps you may see more errors on names and fast speech, but even old dictaphone-quality MP3s usually come out readable.
Can I transcribe a whole podcast episode?
Yes. A 30-minute episode fits the no-signup tier. Typical 45–90 minute episodes fit the free-account tier (up to 1 hour) or Pro (up to 10 hours). Transcribing your back catalog in bulk is what the credit packs are for — $5 covers 10 hours of audio, one-time, no subscription.
Do iPhone and Android voice memos work?
Yes. iPhone Voice Memos exports M4A and most Android recorders export MP3 or M4A — both upload here directly, no conversion needed. Memos recorded in a pocket or across a room will have more errors than ones recorded near the speaker; the editor makes cleanup fast.
Can it transcribe song lyrics from an MP3?
Sometimes, but set expectations: the model is built for speech. Clear, front-and-center vocals often come out largely right; anything with dense instrumentation, stacked harmonies, or rap delivery will be patchy. For lyrics specifically, a dedicated lyrics site usually beats any speech-to-text model.
I have dozens of MP3s — can I do them all?
The free tiers are per-day (3 anonymous / 5 with a free account), so bulk jobs need Pro (fair-use unlimited at $10/month) or a credit pack ($5 for 10 hours, $15 for 40 hours, $39 for 150 hours — packs never expire). Each file is one upload; batch upload is on the roadmap.
Are my ID3 tags or the file itself modified?
No. The MP3 is read, transcribed, and (for anonymous uploads) deleted after 24 hours. We never modify, re-tag, or re-encode the file, and the download you get is the transcript — your original stays exactly as uploaded.