Update app.py

app.py

@@ -7,7 +7,9 @@ import librosa
 import torchaudio
 from scipy.signal import resample
 import time
-from huggingface_hub import snapshot_download, hf_hub_download
+import requests
+
+from huggingface_hub import snapshot_download
 
 ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
 sys.path.append(f'{ROOT_DIR}/third_party/Matcha-TTS')
@@ -15,29 +17,31 @@ sys.path.append(f'{ROOT_DIR}/third_party/Matcha-TTS')
 from cosyvoice.cli.cosyvoice import CosyVoice
 from cosyvoice.utils.file_utils import load_wav
 
-def download_audio_from_dataset(filename, save_path="temp_prompt.wav"):
-    return hf_hub_download(
-        repo_id="kanahomaisa/breezyvoice-samples",
-        filename=filename,
-        repo_type="dataset"
-    )
+def download_audio_from_drive(file_id, save_path="temp_prompt.wav"):
+    url = f"https://drive.google.com/uc?export=download&id={file_id}"
+    response = requests.get(url)
+    if not response.ok or b"<html" in response.content[:100]:
+        raise RuntimeError("Google Drive 音檔下載失敗")
+    with open(save_path, "wb") as f:
+        f.write(response.content)
+    return save_path
 
 preset_speakers = {
     "6歲": {
-        "filename": "smile_train.wav",
+        "file_id": "1PomN05sTjxXBR3q5kOgB7CUEpig2HJmx",
         "transcription": "名字是微笑號，只是呢你們看，這一輛微笑號它這裡有寫八百型的喔，它是八百山出來"
     },
     "8歲": {
-        "filename": "utensils.wav",
+        "file_id": "1U1DKzN_BsU-DL02HZIRlx-opUJLhaQ41",
         "transcription": "出來了出來了，你知道這個餐具是可以挖的，這個餐具可以用窩課魅一起挖嗎?"
     }
 }
 
 def apply_preset(speaker_key):
     if speaker_key in preset_speakers:
-        filename = preset_speakers[speaker_key]["filename"]
+        file_id = preset_speakers[speaker_key]["file_id"]
         transcription = preset_speakers[speaker_key]["transcription"]
-        local_path = download_audio_from_dataset(filename)
+        local_path = download_audio_from_drive(file_id)
         return local_path, transcription
     return None, ""
 
@@ -107,9 +111,8 @@ with gr.Blocks() as demo:
     gr.Markdown("""
     # 小睿語音合成
     選擇預設語音 (6歲、8歲) 或上傳5~15秒語音樣本及該則語音樣本的轉寫，並輸入要合成的句子。
-
     視該時段運算資源和語句長度而定約需70至2000秒，若等候時間過長，可改用較快但較不穩的版本：
-    https://colab.research.google.com/drive/1_p1NLN5tzBwu92ZftPn5vr0U683pTxSY?usp=sharing
+    https://colab.research.google.com/drive/1aQSLwzbK9QnNMH4A5eVfqFYfPFXHwlXx?usp=sharing
     """)
 
     speaker_selector = gr.Dropdown(