tensor

Dockerfile

@@ -8,6 +8,8 @@ RUN pip install --no-cache-dir --upgrade -r /code/requirements.txt
 
 RUN pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cpu
 
+RUN pip install numpy
+
 RUN apt-get -y update
 
 RUN useradd -m -u 1000 user

main.py

@@ -1,7 +1,8 @@
 from flask import Flask, request, jsonify
 from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq
 from transformers import pipeline
-
+import numpy as np
+import json
 app = Flask(__name__)
 model = AutoModelForSpeechSeq2Seq.from_pretrained("GitNazarov/whisper-small-pt-3-uz")
 processor = AutoProcessor.from_pretrained("GitNazarov/whisper-small-pt-3-uz")
@@ -13,7 +14,9 @@ def index():
 @app.route('/transcribe', methods=['POST'])
 def transcribe():
     data = request.json['data']
-    inputs = processor(data, return_tensors="pt", sampling_rate=16000, max_new_tokens=100)
+    data = json.loads(data)
+    tensor_data = np.array(data)
+    inputs = processor(tensor_data, return_tensors="pt", sampling_rate=16000, max_new_tokens=100)
     input_features = inputs.input_features
     generated_ids = model.generate(inputs=input_features)