Create app.py

app.py

@@ -0,0 +1,74 @@
+# app.py
+# simple gradio space for Persian TTS using kamtera/persian-tts-female-vits (coqui tts)
+# all ui messages and comments are in English
+
+import os
+import tempfile
+from hazm import Normalizer
+from TTS.api import TTS
+import gradio as gr
+
+# -------------------------
+# configuration
+MODEL_ID = "Kamtera/persian-tts-female-vits"
+HF_TOKEN = os.environ.get("HF_TOKEN", None)  # optional token for private models
+MAX_INPUT_LENGTH = 1200  # safety limit for long text
+# -------------------------
+
+normalizer = Normalizer()
+
+# load Coqui TTS model
+print("loading tts model:", MODEL_ID)
+if HF_TOKEN:
+    tts = TTS(model_name=MODEL_ID, progress_bar=False, gpu=False, use_auth_token=HF_TOKEN)
+else:
+    tts = TTS(model_name=MODEL_ID, progress_bar=False, gpu=False)
+
+def synthesize(text: str):
+    """
+    text: Persian text input
+    returns: path to the generated wav file
+    """
+    if not text or not text.strip():
+        return None, "please enter some text."
+
+    # limit input length to avoid high latency
+    if len(text) > MAX_INPUT_LENGTH:
+        text = text[:MAX_INPUT_LENGTH] + "."
+
+    # normalize persian text
+    text = normalizer.normalize(text)
+
+    # create a temporary output file
+    out_fd, out_path = tempfile.mkstemp(suffix=".wav")
+    os.close(out_fd)
+
+    # generate audio
+    try:
+        tts.tts_to_file(text=text, file_path=out_path)
+    except Exception as e:
+        print("tts generation error:", e)
+        return None, f"error: {e}"
+
+    return out_path, "speech generated successfully."
+
+# gradio ui
+with gr.Blocks(css=".gradio-container {background-color: #fafafa}") as demo:
+    gr.Markdown("## persian tts — kamtera / persian-tts-female-vits")
+    text_input = gr.Textbox(
+        label="persian text (max ~1200 chars)",
+        lines=6,
+        placeholder="enter your Persian text here..."
+    )
+    generate_btn = gr.Button("generate speech")
+    audio_output = gr.Audio(label="output audio", type="filepath")
+    status = gr.Markdown("")
+
+    def run_tts(text):
+        audio_path, msg = synthesize(text)
+        return audio_path, msg
+
+    generate_btn.click(fn=run_tts, inputs=text_input, outputs=[audio_output, status])
+
+if __name__ == "__main__":
+    demo.launch()