asahi417 commited on
Commit
10d660a
·
verified ·
1 Parent(s): aa89fea

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +24 -2
README.md CHANGED
@@ -18,8 +18,13 @@ This model can be used in CTranslate2 or projects based on CTranslate2 such as [
18
  Install library and download sample audio.
19
  ```shell
20
  pip install faster-whisper
21
- wget https://huggingface.co/kotoba-tech/kotoba-whisper-v1.0-ggml/resolve/main/sample_ja_speech.wav
 
 
 
 
22
  ```
 
23
  Inference with the kotoba-whisper-bilingual-v1.0-faster.
24
 
25
  ```python
@@ -27,9 +32,26 @@ from faster_whisper import WhisperModel
27
 
28
  model = WhisperModel("kotoba-tech/kotoba-whisper-bilingual-v1.0-faster")
29
 
30
- segments, info = model.transcribe("sample_ja_speech.wav", language="ja", chunk_length=15, condition_on_previous_text=False)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
31
  for segment in segments:
32
  print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))
 
33
  ```
34
 
35
  ### Benchmark
 
18
  Install library and download sample audio.
19
  ```shell
20
  pip install faster-whisper
21
+ wget https://huggingface.co/datasets/japanese-asr/en_asr.esb_eval/resolve/main/sample.wav -O sample_en.wav
22
+ wget https://huggingface.co/datasets/japanese-asr/ja_asr.jsut_basic5000/resolve/main/sample.flac -O sample_ja.flac
23
+ ffmpeg -i sample_en.wav -ar 16000 -ac 1 -c:a pcm_s16le sample_en_fixed.wav
24
+ ffmpeg -i sample_ja.flac -ar 16000 -ac 1 -c:a pcm_s16le sample_ja_fixed.wav
25
+
26
  ```
27
+
28
  Inference with the kotoba-whisper-bilingual-v1.0-faster.
29
 
30
  ```python
 
32
 
33
  model = WhisperModel("kotoba-tech/kotoba-whisper-bilingual-v1.0-faster")
34
 
35
+ # Japanese ASR
36
+ segments, info = model.transcribe("sample_ja.flac", language="ja", task="transcribe", chunk_length=15, condition_on_previous_text=False)
37
+ for segment in segments:
38
+ print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))
39
+
40
+ # English ASR
41
+ segments, info = model.transcribe("sample_en_fixed.wav", language="en", task="transcribe", chunk_length=15, condition_on_previous_text=False)
42
+ for segment in segments:
43
+ print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))
44
+
45
+ # Japanese (speech) to English (text) Translation
46
+ segments, info = model.transcribe("sample_ja.flac", language="en", task="translate", chunk_length=15, condition_on_previous_text=False)
47
+ for segment in segments:
48
+ print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))
49
+
50
+ # English (speech) to Japanese (text) Translation
51
+ segments, info = model.transcribe("sample_en.wav", language="ja", task="translate", chunk_length=15, condition_on_previous_text=False)
52
  for segment in segments:
53
  print("[%.2fs -> %.2fs] %s" % (segment.start, segment.end, segment.text))
54
+
55
  ```
56
 
57
  ### Benchmark