Why I'm getting same result with or without using Wav2Vec2Processor?

🤗Tokenizers
1

I’m running simple wav2vec2 code on short without noise voice:

#processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-base-960h")
model     = Wav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-base-960h")     

FILE_NAME        = "tutorial-assets/Lab41-SRI-VOiCES-src-sp0307-ch127535-sg0042.wav"
SPEECH_FILE      = download_asset(FILE_NAME)

speech, sr       = librosa.load(SPEECH_FILE, sr=16000)

speech           = torch.tensor(speech)
speech           = speech.reshape(1, -1)

logits           = model(speech).logits

predicted_ids    = torch.argmax(logits, dim=-1)
transcription    = processor.decode(predicted_ids[0])
transcription

results: 'I HAD THAT CURIOSITY BESIDE ME AT THIS MOMENT'

  • As you see, I didn’t use processor.
  • The examples on the net always used processor

So:

  1. What is the benefit of using processor ?
  2. When do we need to use it ?