Added whisper as a model, time to turn on the GPUs

Dockerfile

@@ -1,5 +1,12 @@
 # 
-FROM python:3.9
+FROM nvidia/cuda:12.4.1-runtime-ubuntu22.04
+
+# Install python and pip
+RUN apt-get update && apt-get install -y python3 python3-pip
+
+# The commands as another prerequisite
+RUN pip install --upgrade pip
+RUN pip install --upgrade git+https://github.com/huggingface/transformers.git accelerate datasets[audio]
 
 # 
 WORKDIR /code

app/whisper.py

@@ -0,0 +1,35 @@
+import torch
+from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
+
+device = "cuda:0" if torch.cuda.is_available() else "cpu"
+torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
+
+model_id = "openai/whisper-large-v3"
+
+model = AutoModelForSpeechSeq2Seq.from_pretrained(
+    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
+)
+model.to(device)
+
+processor = AutoProcessor.from_pretrained(model_id)
+
+pipe = pipeline(
+    "automatic-speech-recognition",
+    model=model,
+    tokenizer=processor.tokenizer,
+    feature_extractor=processor.feature_extractor,
+    max_new_tokens=128,
+    chunk_length_s=30,
+    batch_size=16,
+    return_timestamps=True,
+    torch_dtype=torch_dtype,
+    device=device,
+)
+
+def speech_to_text(path_to_file: str) -> str:
+    """
+    Given the path to the .wav file, it returns the transcription.
+    """
+    result = pipe(path_to_file)
+    
+    return result["text"]

requirements.txt

@@ -1,6 +1,5 @@
 # Add the other requirements here
 fastapi
 uvicorn
-elevenlabs
-mistralai
-streamlit
+torch
+transformers