update app.py

app.py

@@ -42,33 +42,19 @@ VOICE_OPTIONS = [
     "random",  # special option for random voice
 ]
 
-
 def inference(
     text,
-    voice,
-    seed,
+    voice
 ):
     if text is None or text.strip() == "":
-        with open(script.name) as f:
-            text = f.read()
-        if text.strip() == "":
-            raise gr.Error("Please provide either text or script file with content.")
-
-    if split_by_newline == "Yes":
-        texts = list(filter(lambda x: x.strip() != "", text.split("\n")))
-    else:
-        texts = split_and_recombine_text(text)
+        raise gr.Error("Please provide text.")
 
-    voices = [voice]
+    texts = split_and_recombine_text(text)
 
-    if len(voices) == 1:
-        voice_samples, conditioning_latents = load_voice(voice)
-    else:
-        voice_samples, conditioning_latents = load_voices(voices)
+    voice_samples, conditioning_latents = load_voice(voice)
 
     start_time = time.time()
 
-    # all_parts = []
     for j, text in enumerate(texts):
         for audio_frame in tts.tts_with_preset(
             text,
@@ -77,21 +63,21 @@ def inference(
             preset="ultra_fast",
             k=1
         ):
-            # print("Time taken: ", time.time() - start_time)
-            # all_parts.append(audio_frame)
             yield (24000, audio_frame.cpu().detach().numpy())
 
-    # wav = torch.cat(all_parts, dim=0).unsqueeze(0)
-    # print(wav.shape)
-    # torchaudio.save("output.wav", wav.cpu(), 24000)
-    # yield (None, gr.make_waveform(audio="output.wav",))
 def main():
-    title = "Tortoise TTS"
+    title = "Tortoise TTS 🐢"
     description = """
+    A text-to-speech system which powers lot of organizations in Speech synthesis domain.
+    <br/>
+    A model with strong multi-voice capabilities, highly realistic prosody and intonation.
+    <br/>
+    For faster inference, use the 'ultra_fast' preset and duplicate space if you don't want to wait in a queue.
+    <br/>
     """
     text = gr.Textbox(
         lines=4,
-        label="Text (Provide either text, or upload a newline separated text file below):",
+        label="Text:",
     )
 
     voice = gr.Dropdown(
@@ -99,12 +85,12 @@ def main():
     )
 
     output_audio = gr.Audio(label="streaming audio:", streaming=True, autoplay=True)
-    # download_audio = gr.Audio(label="dowanload audio:")
+
     interface = gr.Interface(
         fn=inference,
         inputs=[
             text,
-            voice,
+            voice
         ],
         title=title,
         description=description,
@@ -112,7 +98,6 @@ def main():
     )
     interface.queue().launch()
 
-
 if __name__ == "__main__":
     tts = TextToSpeech(kv_cache=True, use_deepspeed=True, half=True)
 
@@ -121,4 +106,4 @@ if __name__ == "__main__":
             f"\n\n-------------------------Tortoise TTS Scripts Logs, {datetime.now()}-------------------------\n"
         )
 
-    main()
+    main()