4 ms·
What are the best practices for doing the reverse: taking audio and producing text? I don't mind the translation to be rough, the error rate can be quite high f
by ehudla 9y ago
What are the best practices for doing the reverse: taking audio and producing text? I don't mind the translation to be rough, the error rate can be quite high for my purposes, but I want the process not to get stuck and recover so it processes a full length talk.
- Immortalin 9y agoFrom my experiments generating subtitles from TV/movie audiotrack, 75% (worst case) to 95% (best case). If you model it as a standard distribution, somewhere around 85-90% accuracy. Most services provide much better accuracy for stuff like calls or conferences with proper microphones and minimal background noise than for things like TV shows and movies. If the input audio is noisy, I would do some noise filtering before piping it into conversion.
- ehudla 9y agoWhich conversion tools/services do you have in mind?
- Immortalin 9y agoGoogle and Azure