Update app.py

app.py

@@ -3,43 +3,55 @@ import whisper
 import openai
 import os
 
-# Make sure you set your OpenAI API key in the HF Space secrets (Settings → Variables → Add "OPENAI_API_KEY")
+# 🔑 Load OpenAI API key from Hugging Face Space secrets
 openai.api_key = os.environ.get("OPENAI_API_KEY")
 
-# Load Whisper for speech-to-text
+# 🎧 Load Whisper model once (small = faster; use "base" for lighter)
 whisper_model = whisper.load_model("small")
 
-def transcribe_and_translate(audio):
+def transcribe_and_translate(audio_path):
+    if audio_path is None:
+        return "No audio detected.", "No translation."
+
     # Step 1: Transcribe audio → Chinese text
-    result = whisper_model.transcribe(audio, language="zh")
-    chinese_text = result["text"].strip()
+    result = whisper_model.transcribe(audio_path, language="zh")
+    chinese_text = result.get("text", "").strip()
+
+    if not chinese_text:
+        return "⚠️ Could not transcribe audio.", "No translation."
 
     # Step 2: Send transcription to ChatGPT for English translation
     messages = [
-        {"role": "system", "content": "You are a translator. Translate the following Chinese into natural English."},
+        {"role": "system", "content": "You are a professional translator. Translate the following Chinese into fluent, natural English."},
         {"role": "user", "content": chinese_text}
     ]
 
-    response = openai.ChatCompletion.create(
-        model="gpt-4o-mini",  # fast & lightweight, can use gpt-4o if you want best quality
-        messages=messages,
-        temperature=0
-    )
-
-    english_text = response["choices"][0]["message"]["content"].strip()
+    try:
+        response = openai.ChatCompletion.create(
+            model="gpt-4o-mini",  # can change to "gpt-4o" for higher quality
+            messages=messages,
+            temperature=0
+        )
+        english_text = response["choices"][0]["message"]["content"].strip()
+    except Exception as e:
+        english_text = f"⚠️ Translation failed: {e}"
 
     return chinese_text, english_text
 
-# Gradio UI
+# 🚀 Gradio UI
 app = gr.Interface(
     fn=transcribe_and_translate,
-    inputs=gr.Audio(sources=["microphone", "upload"], type="filepath", label="🎙️ Speak Chinese or Upload Audio"),
+    inputs=gr.Audio(
+        sources=["microphone", "upload"],
+        type="filepath",
+        label="🎙️ Speak Chinese or Upload Audio"
+    ),
     outputs=[
-        gr.Textbox(label="🈶 Chinese Text"),
-        gr.Textbox(label="🌍 ChatGPT Translation (English)")
+        gr.Textbox(label="🈶 Chinese Transcription"),
+        gr.Textbox(label="🌍 English Translation (ChatGPT)")
     ],
-    title="🎤 Chinese Voice → English (via ChatGPT)",
-    description="Speak Chinese, the app will transcribe it and then use ChatGPT to translate into English."
+    title="🎤 Chinese Voice → English Translator",
+    description="Speak Chinese or upload audio. Whisper transcribes, then ChatGPT translates into English."
 )
 
 if __name__ == "__main__":