feat: fix bugs

app.py

@@ -1,10 +1,46 @@
-import gradio as gr
 import os
 
+import gradio as gr
+import librosa
+import torch
+from transformers import WhisperForConditionalGeneration, WhisperProcessor
+
 hf_token = os.getenv("hf_token")
 
 if hf_token is None:
-    raise ValueError("Hugging Face token not found. Please set the 'hf_token' environment variable.")
+    raise ValueError(
+        "Hugging Face token not found. Please set the 'hf_token' environment variable."
+    )
+
+processor = WhisperProcessor.from_pretrained(
+    "openai/whisper-small",
+    language="Indonesian",
+    task="transcribe",
+    use_auth_token=hf_token,
+)
+model = WhisperForConditionalGeneration.from_pretrained(
+    "avalonai/whisper-small-jv", use_auth_token=hf_token
+)
+
+
+def transcribe(audio):
+    audio, sampling_rate = librosa.load(audio, sr=16000)
+    audio_input = processor(audio, return_tensors="pt", sampling_rate=16000)
+    input_values = audio_input.input_features
+
+    with torch.no_grad():
+        generated_ids = model.generate(input_values)
+
+    transcription = processor.batch_decode(generated_ids, skip_special_tokens=True)
+    return transcription[0]
+
+
+iface = gr.Interface(
+    fn=transcribe,
+    inputs=gr.Audio(sources="microphone", type="filepath"),
+    outputs="text",
+    title="Speech-to-text on Javanese Language Demo",
+    description="Ini adalah platform untuk pengujian model speech-to-text pada bahasa Jawa oleh Avalon AI. Silahkan coba dengan mengucapkan kalimat",
+)
 
-# Load the model using the access token
-gr.load("models/avalonai/whisper-small-jv", hf_token=hf_token).launch()
+iface.launch()

requirements.txt

@@ -0,0 +1,4 @@
+gradio==4.28.3
+librosa==0.10.1
+torch==2.2.1
+transformers==4.35.2