Whisper model Initialization

app.py

@@ -0,0 +1,55 @@
+import torch
+from transformers import pipeline
+import numpy as np
+import gradio as gr
+
+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-medium"
+
+print("\n\nReading Languages...\n\n")
+
+with open("languages.txt", "r") as file:
+    languages = file.read().strip().split(",")
+
+languages = [language.strip().lower() for language in languages]
+
+print("\n\nInitializing model...\n\n")
+
+transcriber = pipeline(
+    "automatic-speech-recognition",
+    model=model_id,
+    torch_dtype=torch_dtype,
+    device=device,
+)
+
+print("\n\nModel Ready!!\n\nLaunching Interface...\n\n")
+
+def transcribe(audio, language: str):
+    sr, y = audio
+    
+    # Convert to mono if stereo
+    if y.ndim > 1:
+        y = y.mean(axis=1)
+        
+    y = y.astype(np.float32)
+    y /= np.max(np.abs(y))
+    
+    language = language.lower()
+    if(language not in languages):
+        return "Error!! Not a valid language!!"
+
+    args = {"task":"transcribe", "language":language}
+
+    return transcriber({"sampling_rate": sr, "raw": y}, generate_kwargs=args)["text"]  
+
+demo = gr.Interface(
+    transcribe,
+    inputs=[gr.Audio(sources="microphone"), gr.Textbox(label="Language", placeholder="Enter the language")],
+    outputs=["text"],
+    title="Whisper Model Interface",
+    description=model_id
+)
+
+demo.launch()

languages.txt

@@ -0,0 +1 @@
+Afrikaans, Arabic, Armenian, Azerbaijani, Belarusian, Bosnian, Bulgarian, Catalan, Chinese, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, Galician, German, Greek, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Italian, Japanese, Kannada, Kazakh, Korean, Latvian, Lithuanian, Macedonian, Malay, Marathi, Maori, Nepali, Norwegian, Persian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swahili, Swedish, Tagalog, Tamil, Thai, Turkish, Ukrainian, Urdu, Vietnamese, Welsh

requirements.txt

@@ -0,0 +1,7 @@
+gradio
+transformers[torch]
+torchaudio
+sentencepiece
+tiktoken
+accelerate
+numpy