Migrated to end-to-end models trained using Icefall

README.md

@@ -12,13 +12,14 @@ It consists of the following components:
 
  * Speech activity detector (https://github.com/snakers4/silero-vad)
  * Online speaker change detector (https://github.com/alumae/online_speaker_change_detector)
- * Speech recognition: we use a forked version of Vosk API (https://github.com/alphacep/vosk-api), which is in turn based on Kaldi
- * Unknown word reconstuctor (words not in the speech recognition vocabulary are reconstructed using a phoneme-to-grapheme tool, which is based on FSTs and an n-gram model -- https://github.com/alumae/et-g2p-fst)
- * Compound word recognizer: for glueing together compound word tokens, using an n-gram model
- * Puncutator, LSTM-based (https://github.com/alumae/streaming-punctuator)
- * Words-to-numbers converter (FST-based, using Pynini)
+ * Speech recognition: we use streaming transducer models trained using [Icefall](https://github.com/k2-fsa/icefall). Speech recognition model
+ includes punctuation insertion. [Sherpa-ONNX](https://github.com/k2-fsa/sherpa-onnx) is used for decoding.
+
 
 ## News
+ * 2023-03-30: *Major rewrite*: Now uses end-to-end streaming transducer models trained using [Icefall](https://github.com/k2-fsa/icefall).
+ The model includes punctuation restoration. Should provide 30% less ASR errors and much improved punctuation accuracy.
+ Also, the Docker image size reduced to from ~14 GB to 3.2 GB, as we don't need Kaldi any more.
  * 2023-01-01: Fixed a bug in Youtube presenter that caused timings to be sometimes messed up. Also introduced
  a new command line argument `--word-output-delay <seconds>' that can be used to set a minimum time between a word
  occurring in speech and word appearing in subtitles. Recommended value is smth like 5. This makes word-by-word output

main.py

@@ -86,7 +86,7 @@ def main(args):
  encoder="models/sherpa/encoder.onnx",
  decoder="models/sherpa/decoder.onnx",
  joiner="models/sherpa/joiner.onnx",
- num_threads=4,
+ num_threads=2,
  sample_rate=16000,
  feature_dim=80,
  enable_endpoint_detection=True,