Update README.md

README.md

@@ -1,6 +1,6 @@
 # Unsupervised malay speakers from youtube videos
 
-10492 unique speakers with at least 75 hours of speech. Steps to reproduce at https://github.com/huseinzol05/malaya-speech/blob/master/data/youtube/process-youtube.ipynb
+10492 unique speakers with at least 75 hours of voice activities. Steps to reproduce at https://github.com/huseinzol05/malaya-speech/blob/master/data/youtube/process-youtube.ipynb
 
 ## how-to
 
@@ -68,8 +68,8 @@
 
 - all mp3 files postprocessing using https://malaya-speech.readthedocs.io/en/latest/load-noise-reduction.html and https://malaya-speech.readthedocs.io/en/latest/load-speech-enhancement.html
 - `wav_data` is directory of the audio, prune the path to proper extracted directory.
-- `asr_model` is predicted using the best model that we have, `conformer-medium`, returned `(text, confidence, subwords)`.
-- `classification_model` is predicted using NEMO TITANET Large speaker verification model, https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/titanet_large.
+- `asr_model` is predicted using the best model that we have, `conformer-medium`, returned `(text, probability, subwords)`.
+- `classification_model` is predicted using NEMO TITANET Large speaker verification model, https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/titanet_large, with streaming speaker similarity, https://malaya-speech.readthedocs.io/en/latest/huggingface-repository.html
 
 3. Group by similar speakers using pagerank method (scipy.sparse.linalg.gmres),