This reads the word-level alignment data that belongs to a song generated on this site. It is not speech recognition run over arbitrary audio: there is no upload field, and a file from your disk has no alignment data to look up. If you need timings for a recording you made elsewhere, this is the wrong tool.
Timestamped Lyrics with Word-Level Timing
Get word-level timings for the lyrics of a song you generated on this site. Pick a finished vocal track, check the alignment, and export a timed text file.
Pick a song you generated on this site and get its lyrics back with a start and end time on every word. Play the track, watch the current word highlight, click any word to jump there, and export the timings as text.
01
What This Tool Aligns, and What It Does Not
What comes back is a list of words, each with a start time, an end time and a flag saying whether that particular word aligned successfully. Alignment is per word rather than per line, which is a meaningful difference: a line-level tool tells you when a line begins, while word-level timing tells you where you are inside it.
Because the data is tied to the generated track, the lyric it aligns is the lyric the song was generated from. You are not transcribing a performance and hoping the words come out right; the words are already known, and the only question being answered is when each one is sung.
02
Getting the Timings
Sign in and open the tool, then choose a finished song with vocals from your music history. Only completed generations appear, because a track still rendering has no audio to align against. Each request costs 5 credits.
Instrumental generations are the usual reason a request returns nothing. There are no words to place, so the result comes back empty rather than wrong — if you get an empty result, check that the track you picked actually has a vocal.
Once the timings arrive, the lyric appears beside the player. The word matching the current playback position is highlighted as the song runs, and clicking any word seeks the player to that word's start time. That click-to-seek behaviour is the fastest way to check a section without scrubbing the waveform by hand.
03
Why the Alignment Is Worth Checking
Each word carries a success flag, and not every word gets one. Words that failed to align still appear in the list so the lyric stays readable, but their timing should not be trusted without listening. Scanning for them first is quicker than reading the whole lyric against the audio.
Three places go wrong more often than the rest. A long held note spread across several syllables tends to collect its words at the start of the phrase rather than spreading them across it. Rapid passages pack words so tightly that a neighbouring word can take the position. And ad-libs or doubled vocals sitting behind the lead can pull a timing toward the background line.
Everything else is usually close enough to use as-is. The practical approach is to play the track once from the top, watch where the highlight drifts out of step with what you hear, and only inspect those moments.
04
The Exported File
The export is a plain text file with one word per line, written as a timestamp in square brackets followed by the word itself. A line looks like a minute and second marker and then a single word, which makes the file easy to read and easy to parse.
Two details matter if you plan to feed it into something else. Timestamps are rounded down to whole seconds, so several consecutive words often share the same marker in fast passages. And because each line holds one word rather than one lyric line, converting to a line-based format means grouping words yourself.
Subtitle and karaoke formats are not exported directly. Building an LRC or SRT file from this data is straightforward if you need one, but it involves deciding where lines break and how long each line should stay on screen, and those are choices worth making deliberately rather than having a tool guess them.
05
Where It Fits Among the Other Tools
The boundary is the source. This tool only works on songs generated on this site, because the alignment data is produced alongside the generation and stored with the track. Nothing here applies to an MP3 or WAV you bring in from outside.
If you have your own recording and want words placed against it, that is transcription rather than alignment, and it is a different job with different accuracy. If what you want is the words themselves rather than their timing, the lyrics generator writes a lyric from a theme, and the music generator turns a lyric into a sung track.
If you are working with an instrumental and want its structure rather than its words, the key and tempo finder reads tempo and key from the audio, which is usually what people are actually after when they ask for timing information about a song without vocals.
06
Timestamped Lyrics Questions
Can I upload my own audio file?
No. The tool reads alignment data stored with a song generated on this site, so there is nothing to look up for an outside file. There is no upload field for that reason.
Are the timings per line or per word?
Per word. Every word comes back with its own start and end time, which is what lets the viewer highlight the exact word playing and seek to it on a click.
How many credits does one request cost?
Five credits per request, charged when you submit a finished vocal track from your music history.
Why did my request come back empty?
Almost always because the track has no vocal. An instrumental generation has no words to place, so the result is empty rather than approximate. Pick a generation that sings.
What format is the download?
A plain text file with one word per line, each preceded by its timestamp in square brackets. Timestamps are rounded down to the nearest second.
Can I export LRC or SRT subtitles?
Not directly. The text export holds the same timings, but turning it into a subtitle file means grouping words into lines and choosing how long each line shows, which the tool leaves to you.
Some words look like they are timed slightly early. Why?
Held notes and fast passages are the usual causes, and each word carries a flag showing whether it aligned successfully. Play the section and adjust those moments rather than the whole lyric.
Does it work for any language the song is in?
Alignment follows the lyric the song was generated from, so it is not limited to English. Accuracy still tracks how clearly the vocal is sung, in any language.