Why make an SRT from audio?
- Edit captions before the video exists. Export the voice-over from your editor, caption it here, and drop the SRT back onto the timeline; it lines up because the timing starts at zero in both.
- Audiograms and podcast clips. Social clips of a podcast need captions to be watched on mute. Generate the SRT here, then burn it into the clip once you have the video.
- Lyrics. The transcription model can pick up sung vocals reasonably well when the mix is not too dense; the result is a good start for a lyric video or LRC file, but expect to correct words.
- Accessible audio. A timestamped transcript lets deaf and hard-of-hearing listeners follow an episode, and search engines index it.
Getting clean timing from audio
The transcription model returns a timestamp roughly every sentence. SubtitleWave splits long sentences at natural breaks and shares the time across the parts by length, which is accurate to within a few tenths of a second for normal speech. For music, rhythmic speech or long pauses, check the timeline: every subtitle is a black block over the waveform, and the waveform peaks show exactly where words start.
Questions people ask
Which audio formats are supported?
Anything your browser can play: MP3, WAV, M4A/AAC, OGG/Opus, FLAC and WebM audio are supported in all modern browsers. Apple CAF and WMA usually are not.
Can I make an SRT from a song?
Yes, but accuracy on sung vocals is lower than on speech, especially with loud instruments. Set the spoken language, then correct the lines in the editor.
Can I get a VTT or TXT instead?
Yes. The Export tab offers SRT, VTT, SBV, ASS and plain text (with or without timestamps) from the same result.