UPDATE: You can now change the audio speed via the AA tab in the sidebar:
2 Likes
I just want to chime in here and say that I just upgraded to Readlang Premium last night and have been experimenting with these features. You are definitely on to something great by adding this in. It seems to work better than the old sync method of watching through the video and clicking words, so I am all for it.
I use Whisper locally to make a transcript with a fine tuned French model and then have Ollama use a local LLM to grammar correct transcripts before force aligning them with Whisper again to get word accurate timings of the optimized transcripts. This process outputs a basic text file with paragraph breaks (that looks good for reading view) as well as an SRT file with word-by-word, phrase or sentence timings that have been timed correctly for the audio I have provided. If possible, I would like a way to upload these three files when adding new items to my library so that you don’t need to pay for whisper processing time as I already have the files generated locally.
If you want examples or more details about how I achieve this locally on my mac just let me know. I use some free command line tools paired with Hazel to automate it whenever a new mp3 gets placed in a certain directory and it produces the appropriate output. The quality of these transcripts seems higher than the auto-generated ones YouTube produces from what I can tell.
Keep up the good work! Thanks again.
1 Like
This is one of the reason’s I use Lingq as well as Readlang: I upload my own .srt transcript file along with the audio to create Lingq “lessons.” Lingq reader’s “sentence” mode is invaluable to play the audio for any given “sentence” to double check the transcription and edit the “sentence” or simply to listen to the pronunciation. Great for study.
1 Like
This!!! Sentence mode + audio/text synch are the two best LingQ features readlang lacks atm
1 Like