Delete pipeline.py

pipeline.py

@@ -1,46 +0,0 @@
-import numpy as np
-from typing import Dict
-
-from transformers import  Wav2Vec2Processor, Wav2Vec2ForCTC
-from pyctcdecode import Alphabet, BeamSearchDecoderCTC
-
-class PreTrainedPipeline():
-    def __init__(self, path):
-        """
-        Initialize model
-        """
-        self.processor = Wav2Vec2Processor.from_pretrained(path)
-        self.model = Wav2Vec2ForCTC.from_pretrained(path)
-        vocab_list = list(self.processor.tokenizer.get_vocab().keys())
-
-        # convert ctc blank character representation
-        vocab_list[0] = ""
-
-        # replace special characters
-        vocab_list[1] = "⁇"
-        vocab_list[2] = "⁇"
-        vocab_list[3] = "⁇"
-
-        # convert space character representation
-        vocab_list[4] = " "
-
-        alphabet = Alphabet.build_alphabet(vocab_list, ctc_token_idx=0)
-
-        self.decoder = BeamSearchDecoderCTC(alphabet)
-        self.sampling_rate = 16000
-
-
-    def __call__(self, inputs)-> Dict[str, str]:
-        """
-        Args:
-            inputs (:obj:`np.array`):
-                The raw waveform of audio received. By default at 16KHz.
-        Return:
-            A :obj:`dict`:. The object return should be liked {"text": "XXX"} containing
-            the detected text from the input audio.
-        """
-        input_values = self.processor(inputs, return_tensors="pt", sampling_rate=self.sampling_rate).input_values  # Batch size 1
-        logits = self.model(input_values).logits.cpu().detach().numpy()[0]
-        return {
-            "text": self.decoder.decode(logits)
-        }