Upload 2 files

app.py

@@ -0,0 +1,23 @@
+import gradio as gr
+from transformers import pipeline
+import torch
+
+# モデルのロード (例: Tacotron2 + WaveGlow)
+synthesizer = pipeline("text-to-speech", "tts_model_name", device=0)  # モデル名を指定, GPUを使う場合は device=0
+
+# 推論関数の定義
+def synthesize_speech(text):
+    with torch.no_grad():
+      audio = synthesizer(text)
+    return (synthesizer.sampling_rate, audio["audio"].numpy())
+
+# Gradioインターフェースの作成
+iface = gr.Interface(
+    fn=synthesize_speech,
+    inputs=gr.Textbox(lines=5, label="Input Text"),
+    outputs=gr.Audio(label="Output Audio"),
+    title="TTS Demo",
+    description="Enter text to synthesize speech."
+)
+
+iface.launch()

requirements.txt

@@ -0,0 +1,4 @@
+gradio
+transformers
+torch
+espnet