SaarAI commited on
Commit
d67bf19
·
verified ·
1 Parent(s): dae4434

Model card: public release

Browse files
Files changed (1) hide show
  1. README.md +2 -2
README.md CHANGED
@@ -36,7 +36,7 @@ tags:
36
 
37
  # Amharic-ASR-v2
38
 
39
- > **Private release.** Benchmarked on the full SaarAI Amharic ASR leaderboard
40
  > (9 test sets). v2 + LM has the **best average WER (27.15%)** of all models
41
  > tested, ahead of v1 (28.19%) and `addisai/addis-scribe-streaming` (31.62%).
42
 
@@ -77,7 +77,7 @@ pip install transformers torch torchaudio pyctcdecode kenlm # kenlm needs Pyth
77
  from transformers import pipeline
78
 
79
  asr = pipeline("automatic-speech-recognition",
80
- model="SaarAI/Amharic-ASR-v2", # private: log in with `hf auth login`
81
  device=0)
82
  print(asr("audio.wav")["text"]) # uses the LM
83
  print(asr("call.wav", chunk_length_s=20, stride_length_s=4)["text"]) # long audio
 
36
 
37
  # Amharic-ASR-v2
38
 
39
+ > Benchmarked on the full SaarAI Amharic ASR leaderboard
40
  > (9 test sets). v2 + LM has the **best average WER (27.15%)** of all models
41
  > tested, ahead of v1 (28.19%) and `addisai/addis-scribe-streaming` (31.62%).
42
 
 
77
  from transformers import pipeline
78
 
79
  asr = pipeline("automatic-speech-recognition",
80
+ model="SaarAI/Amharic-ASR-v2",
81
  device=0)
82
  print(asr("audio.wav")["text"]) # uses the LM
83
  print(asr("call.wav", chunk_length_s=20, stride_length_s=4)["text"]) # long audio