Update app.py

app.py

@@ -1,61 +1,21 @@
-import streamlit as st
-import os
-import base64
-import uuid
 
-st.title("Record Audio in Browser")
-
-# JavaScript to record audio
-audio_recorder_js = """
-<script>
-let mediaRecorder;
-let audioChunks = [];
-
-function startRecording() {
-    navigator.mediaDevices.getUserMedia({ audio: true })
-        .then(stream => {
-            mediaRecorder = new MediaRecorder(stream);
-            mediaRecorder.ondataavailable = event => {
-                audioChunks.push(event.data);
-            };
-            mediaRecorder.onstop = () => {
-                const audioBlob = new Blob(audioChunks, { type: 'audio/wav' });
-                const reader = new FileReader();
-                reader.readAsDataURL(audioBlob);
-                reader.onloadend = () => {
-                    const base64AudioMessage = reader.result.split(',')[1];
-                    fetch('/save_audio', {
-                        method: 'POST',
-                        body: JSON.stringify({ audio: base64AudioMessage }),
-                        headers: { 'Content-Type': 'application/json' }
-                    }).then(response => response.json()).then(data => {
-                        console.log(data);
-                    });
-                };
-            };
-            mediaRecorder.start();
-        });
-}
-
-function stopRecording() {
-    mediaRecorder.stop();
-}
-</script>
-
-<button onclick="startRecording()">Start Recording</button>
-<button onclick="stopRecording()">Stop Recording</button>
-"""
-
-st.components.v1.html(audio_recorder_js)
-
-# Backend to save audio
-if "audio_data" not in st.session_state:
-    st.session_state["audio_data"] = None
-
-if st.session_state["audio_data"]:
-    audio_bytes = base64.b64decode(st.session_state["audio_data"])
-    file_name = f"recording_{uuid.uuid4()}.wav"
-    with open(file_name, "wb") as f:
-        f.write(audio_bytes)
-    st.audio(file_name, format="audio/wav")
-    st.success(f"Audio saved as {file_name}")
+# Setup model
+device = "cuda:0" if torch.cuda.is_available() else "cpu"
+torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
+model_id = "KBLab/kb-whisper-tiny"
+
+@st.cache_resource
+def load_model():
+    model = AutoModelForSpeechSeq2Seq.from_pretrained(
+        model_id, torch_dtype=torch_dtype, use_safetensors=True, cache_dir="cache"
+    )
+    model.to(device)
+    processor = AutoProcessor.from_pretrained(model_id)
+    return pipeline(
+        "automatic-speech-recognition",
+        model=model,
+        tokenizer=processor.tokenizer,
+        feature_extractor=processor.feature_extractor,
+        torch_dtype=torch_dtype,
+        device=device,
+    )