MANI2003 commited on
Commit
b54f577
Β·
verified Β·
1 Parent(s): 50d8efe

Delete AT.py

Browse files
Files changed (1) hide show
  1. AT.py +0 -55
AT.py DELETED
@@ -1,55 +0,0 @@
1
- import torch
2
- import torchaudio
3
- from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer
4
- import gradio as gr
5
-
6
- # 1️⃣ Load Whisper model for Speech-to-Text
7
- asr_pipe = pipeline("automatic-speech-recognition", model="openai/whisper-small")
8
-
9
- # 2️⃣ Load Qwen-style LLM for text response
10
- model_name = "Qwen/Qwen1.5-0.5B-Chat"
11
- tokenizer = AutoTokenizer.from_pretrained(model_name)
12
- model = AutoModelForCausalLM.from_pretrained(model_name, device_map="auto",torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32)
13
-
14
- # Bot reply generator
15
- def generate_response(user_text):
16
- messages = [
17
- {"role": "system", "content": "You are a helpful assistant."},
18
- {"role": "user", "content": user_text}
19
- ]
20
-
21
- # Use chat template formatting
22
- input_ids = tokenizer.apply_chat_template(messages, return_tensors="pt").to(model.device)
23
-
24
- with torch.no_grad():
25
- output = model.generate(input_ids, max_new_tokens=150, pad_token_id=tokenizer.eos_token_id)
26
-
27
- decoded = tokenizer.decode(output[0], skip_special_tokens=True)
28
-
29
- # Return only the assistant's message
30
- response = decoded.split("assistant")[-1].strip().replace(":", "").strip()
31
-
32
- return response
33
-
34
-
35
- # Complete pipeline: Audio β†’ Text β†’ Response
36
- def audio_to_bot_response(audio_path):
37
- print("[INFO] Transcribing audio...")
38
- result = asr_pipe(audio_path)
39
- user_text = result['text']
40
-
41
-
42
- print(f"[INFO] Transcribed: {user_text}")
43
- response = generate_response(user_text)
44
-
45
- return f"πŸ‘€ You said: {user_text}\nπŸ€– Bot: {response}"
46
-
47
- interface = gr.Interface(
48
- fn=audio_to_bot_response,
49
- inputs=gr.Audio(sources=["microphone"], type="filepath"),
50
- outputs="text",
51
- title="πŸŽ™ Voice to AI Bot Response",
52
- description="Speak into the mic. The AI will transcribe and respond."
53
- )
54
-
55
- interface.launch()