Add C++ runtime and Python APIs for Moonshine models

[csukuangfj]

Note that Moonshine tiny is much faster than whisper tiny.en.

RTF on my MacBook Pro (CPU, 1 thread)

	Moonshine tiny	Whisper tiny.en
RTF	0.031	0.095

Speed test about generating subtitles (1 thread, CPU, on my MacBook Pro)

Please first download model files from https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models

wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/Obama.wav

wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-moonshine-tiny-en-int8.tar.bz2
tar xf sherpa-onnx-moonshine-tiny-en-int8.tar.bz2

wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/sherpa-onnx-whisper-tiny.en.tar.bz2
tar xf sherpa-onnx-whisper-tiny.en.tar.bz2

Moonshine tiny

./python-api-examples/generate-subtitles.py  \
  --silero-vad-model=./silero_vad.onnx \
  --moonshine-preprocessor=./sherpa-onnx-moonshine-tiny-en-int8/preprocess.onnx \
  --moonshine-encoder=./sherpa-onnx-moonshine-tiny-en-int8/encode.int8.onnx \
  --moonshine-uncached-decoder=./sherpa-onnx-moonshine-tiny-en-int8/uncached_decode.int8.onnx \
  --moonshine-cached-decoder=./sherpa-onnx-moonshine-tiny-en-int8/cached_decode.int8.onnx \
  --tokens=./sherpa-onnx-moonshine-tiny-en-int8/tokens.txt \
  --num-threads=1 \
  /Users/fangjun/Desktop/Obama.mov

Outputs

Started!
Saved to /Users/fangjun/Desktop/Obama.srt
Audio duration: 335.235 s
Elapsed:        10.464 s
RTF = 10.464/335.235 = 0.031
Done!

Whisper tiny.en

./python-api-examples/generate-subtitles.py  \
  --silero-vad-model=./silero_vad.onnx \
  --whisper-encoder=./sherpa-onnx-whisper-tiny.en/tiny.en-encoder.int8.onnx \
  --whisper-decoder=./sherpa-onnx-whisper-tiny.en/tiny.en-decoder.int8.onnx \
  --tokens=./sherpa-onnx-whisper-tiny.en/tiny.en-tokens.txt \
  --num-threads=1 \
  /Users/fangjun/Desktop/Obama.mov

Output

Started!
Saved to /Users/fangjun/Desktop/Obama.srt
Audio duration: 335.235 s
Elapsed:        31.809 s
RTF = 31.809/335.235 = 0.095
Done!

Config of my MacBook Pro