Create app.py

app.py

@@ -0,0 +1,29 @@
+import gradio as gr
+import numpy as np
+import torch
+from transformers import pipeline, AutoModelForSpeechSeq2Seq, AutoProcessor
+
+model_id = 'openai/whisper-large-v3'
+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).to(device)
+processor = AutoProcessor.from_pretrained(model_id)
+
+pipe_asr = pipeline("automatic-speech-recognition", model=model, tokenizer=processor.tokenizer, feature_extractor=processor.feature_extractor, max_new_tokens=128, chunk_length_s=15, batch_size=16, torch_dtype=torch_dtype, device=device, return_timestamps=True)
+
+base_audio_drive = "/data/audio"
+
+def transcribe_function(audio):
+    sr, y = audio[0], audio[1]
+    y = y.astype(np.float32) / np.max(np.abs(y))
+    result = pipe_asr({"array": y, "sampling_rate": sr}, return_timestamps=False)
+    full_text = result.get("text", "")
+    return full_text
+
+def gradio_transcribe(audio):
+    sr, y = audio
+    return transcribe_function((sr, y))
+
+iface = gr.Interface(fn=gradio_transcribe, inputs=gr.inputs.Audio(source="microphone", type="numpy"), outputs="text", title="Voice to Text Transcription", description="Transcribe your voice input to text using a pre-trained Whisper model.")
+
+iface.launch()