Use cases

Song Lyrics to Text from a Recording

Upload a song or a vocal recording and get a timestamped draft of the lyrics to correct along with the player. Speech recognition is built for talking, so expect more corrections than with speech.

  • Free, no sign-up
  • No credit card
  • 90+ languages
  • TXT, DOCX, SRT, VTT, PDF

New transcription

Try one file now, no sign-up needed. A free account keeps your transcripts and lets you do more.
Auto-detected by default
Label who said what
A last resort for very noisy recordings; most files are better without it

How it works

Three steps, no setup.

  1. 01Step · 01

    Upload

    Drop an audio or video file. Choose the language and whether to label speakers.

  2. 02Step · 02

    Transcribe

    We extract the audio, transcribe it and identify speakers. Watch progress live.

  3. 03Step · 03

    Use it

    Read along with the player, jump by clicking a line, and share the transcript.

Set expectations: music lowers accuracy

The model behind RocketScribe was trained mostly on speech. Singing stretches vowels, bends pitch and sits under drums, guitars and synths, so a full mix transcribes noticeably worse than a podcast. Dense mixes, heavy effects on the voice, fast rap and backing vocals singing different words all make it harder.

Instrumental passages can produce stray words or a repeated line where nothing was sung, and a chorus can be dropped or duplicated. Treat the output as a draft that saves you typing, then correct it by ear.

What gives the best result

Clear vocals are what matters most:

  • An isolated vocal track or stem from your own session gives far better results than the final mix.
  • Acoustic recordings with one voice and one instrument come out better than full-band arrangements.
  • A demo sung on a phone in a quiet room works well for songwriters capturing ideas.
  • Pick the language of the song rather than auto-detect, especially when the intro is instrumental.

Where songs come from

Upload MP3, M4A, WAV, FLAC or OGG audio, or a music video as MP4 or MOV, as it is. A public YouTube, TikTok or SoundCloud link can be imported with Import from link. Only the audio is fetched. Streaming services do not give you a file to upload, so this works with recordings you have.

Songwriters often use the browser mic recording in the upload box to sing a verse straight in and get the words down while the idea is fresh.

Correcting and exporting lyrics

The current line is highlighted as the song plays. Click a line to replay it and edit the words in place. Line breaks follow the timing of the recording, so you may want to rearrange them into verses and choruses in your own document.

Export TXT or press Copy for a lyric sheet. SRT or VTT give timed lines you can use as captions on your own lyric video or performance clip; subtitles are not burned into the video. DOCX suits a songbook or a lyric sheet for a choir or band.

Copyright

Lyrics are usually protected by copyright. Transcribing your own songs, songs you have rights to, or lyrics for private study and practice is one thing; publishing someone else's lyrics or captioned videos of their songs generally needs permission from the rights holder. For songs on YouTube, see the YouTube song lyrics transcript page.

FAQ

Song Lyrics to Text: questions, answered.

How accurate is lyric transcription?+

Lower than for speech. Clear, isolated vocals give a usable draft; a busy full mix needs many corrections.

Why are there words where nobody sings?+

Speech models sometimes write stray words during instrumental parts. Delete them in the editor.

Can I make a lyric video subtitle file?+

Yes. Export SRT or VTT and add it as captions to your own video. Subtitles are not burned in.

Can I transcribe a song from Spotify or Apple Music?+

No. Streaming services do not provide a file or link we can import. Use a recording you own, or a public YouTube, TikTok or SoundCloud link.

Does it translate lyrics?+

No. Lyrics are written in the language that was sung. A Google Translate link is available on the transcript.

Do I need an account?+

No. You can transcribe one file right away without signing up. A free account then gives you 3 files a day (120 minutes a month) and keeps your transcripts: the file you transcribed before signing up moves into the account.

Which formats are supported?+

Any common audio or video file: MP3, WAV, M4A, AAC, FLAC, OGG, MP4, MOV, WEBM, MKV and more. We extract the audio from video automatically.

How accurate is it?+

On clear recordings expect accuracy comparable to leading transcription services. Background noise, crosstalk and heavy accents reduce it, as they do everywhere.

Pricing

Start free. Pay only if you need more.

The Free plan needs no card. Paid plans add longer files, batch uploads and speaker identification.

Free

$0

no card required

For trying it out and the occasional file.

  • 3 files a day
  • 120 minutes a month
  • Timestamps and synced player
  • AI prompt export and translation
  • Transcripts kept for 30 days
Start free
PlusMost popular

$6/ month

billed $72 yearly

For people who transcribe every week.

  • 1,200 minutes of transcription a month
  • Speaker identification included
  • Files up to 300 minutes and 5 GB
  • Upload 20 files at once
  • Up to 5 files processed in parallel
  • Priority processing
Get Plus
Unlimited

$10/ month

billed $120 yearly

For heavy users and small teams.

  • Unlimited transcription*
  • Speaker identification, 2,400 minutes a month
  • Files up to 600 minutes and 5 GB
  • Upload 50 files at once
  • Up to 5 files processed in parallel
  • Priority processing
Get Unlimited

Minute Packs

One-timefrom $1 per 60 minutes
  • Plus limits on Free: longer files, batch uploads
  • Speaker identification
  • No RocketScribe line in your exports
  • Valid 12 months, packs stack

Top up any plan. Stackable and valid for 12 months. A pack's minutes are one shared pool for whatever your plan doesn't cover: transcription on Free and Plus, speaker identification (and transcription past fair use) on Unlimited. On Free, a pack also unlocks Plus limits, with transcription and speaker identification both drawn from it.

* Fair use: 12,000 minutes a month. Beyond that, uploads pause until the next cycle or a Minute Pack. Prices in USD.

Song Lyrics to Text from a Recording

Pick a file, choose the language, press Transcribe. No sign-up needed to try.

Transcribe a file