Upload gradio_app_transcribe.py

gradio_app_transcribe.py

@@ -0,0 +1,54 @@
+import gradio as gr
+from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
+import torch
+import requests
+import config
+import os
+
+def fetch_access_token():
+    token_response = requests.post(token_url, timeout=15)
+    token_response.raise_for_status()
+    token = token_response.json()
+    return token["access_token"]
+
+client_id = config.client_id
+client_secret = config.client_secret
+token_url = "https://id.twitch.tv/oauth2/token?client_id=" + client_id + "&client_secret=" + client_secret + "&grant_type=client_credentials"
+model_id = "distil-whisper/distil-large-v2"
+
+access_token = fetch_access_token()
+
+device = "cuda:0" if torch.cuda.is_available() else "cpu"
+torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
+
+model = AutoModelForSpeechSeq2Seq.from_pretrained(
+    model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
+)
+model.to(device)
+
+torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
+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,
+    torch_dtype=torch_dtype,
+    device=device,
+)
+
+def transcribe_audio(audio_file):
+    recorded_filename = audio_file.name
+    if os.path.exists(recorded_filename):
+        results = pipe(recorded_filename)
+        return results["text"]
+    else:
+        return "Error: No audio file uploaded."
+
+inputs = gr.Audio(sources="upload", type="filepath")
+outputs = gr.Textbox()
+
+interface = gr.Interface(fn=transcribe_audio, inputs=inputs, outputs=outputs, title="Audio Transcription App")
+interface.launch()