issues
search
princeton-ddss
/
SpeechMLPipeline
SpeechMLPipeline is a complete pipeline to deploy Machine Learning Models to generate labelled and timestamped transcripts from audio inputs
MIT License
0
stars
1
forks
source link
Apply llama2 to Speaker Diarization
#13
Closed
fjying
closed
7 months ago
fjying
commented
1 year ago
[x] Read llama2 Documentation
[x] Download llama2 Model
[x] Build Sample Codes of Applying llama2 Model to Diarization
[x] Run LLaMA Model on Della using GPU
[x] Prompting to ensure that the model gives correct outputs in a standardized format
[x] Parse Whisper Outputs into Question in the llama2 prompt
[x] Need to separate and pass segments of Whisper Outputs into Question to not exceed the maximum length of prompt
[ ] Prompting to Improve the Accuracy of Segmentation and Diarization #16
[x] Parse llama2 diarization outputs into pandas dataframe
[ ] Merge llama2 diarization output files with whisper output file