Standard Haitian orthography and chunking
#2
by lmselby - opened
Hello, thank you for sharing your model. It is very difficult to find good ASR for Haitian. I am new to Hugging Face.
I am comparing various ASR approaches I can run in Colaboratory with other types of transcriptions. I am noticing that the results from running 30 second .wav files on your model results in some standard Haitian orthography, some approximation of the sound that does not conform to standard orthography. Could you comment on this challenge, please?
I have a nearly 21-minute, 19 MB .mp3 that I would like to transcribe without manually cutting it into clips in audio-editing software. Could you share the code I would necessarily need to add for chunking the audio to run with your model? Thank you.