Update app.py

app.py

@@ -2,7 +2,7 @@ import torch
 import os
 import random
 import gradio as gr
-
+from TTS.api import TTS
 from transformers import SpeechT5Processor, SpeechT5ForTextToSpeech, SpeechT5HifiGan, pipeline
 import base64
 from datasets import load_dataset
@@ -26,18 +26,13 @@ def guessanAge(model, image):
     return description    
 
 @spaces.GPU(duration=120)
-def text2speech(model, text, voice):
+def text2speech(text, sample):
+    print(TTS().list_models())
     print(voice)
     if len(text) > 0:
-        synthesiser = pipeline("text-to-speech", model=model)
-        
-        embeddings_dataset = load_dataset("Matthijs/cmu-arctic-xvectors", split="validation")
-        speaker_embedding = torch.tensor(embeddings_dataset[voice]["xvector"]).unsqueeze(0)
-        
-        speech = synthesiser(text, forward_params={"speaker_embeddings": speaker_embedding})
-        audio_data = np.frombuffer(speech["audio"], dtype=np.float32)
-        audio_data_16bit = (audio_data * 32767).astype(np.int16)
-        return speech["sampling_rate"], audio_data_16bit
+        tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2")
+        wav = tts.tts(text="Hello world!", speaker_wav=sample, language="en")
+        return wav
 
 @spaces.GPU
 def ImageGenFromText(text, model):
@@ -77,11 +72,9 @@ tab2 = gr.Interface(
     outputs=["text"],
 )
 textbox = gr.Textbox(value="good morning pineapple! looking very good very nice!")
-radio3 = gr.Radio(["microsoft/speecht5_tts"], value="microsoft/speecht5_tts", label="Select an tts", info="Age Classifier")
-radio3_1 = gr.Radio([("Scottish male (awb)", 0), ("US male (bdl)", 1138), ("US female (clb)", 2271), ("Canadian male (jmk)",3403), ("Indian male (ksp)", 4535), ("US male (rms)", 5667), ("US female (slt)", 6799)], value=4535)
 tab3 = gr.Interface(
     fn=text2speech,
-    inputs=[radio3, textbox, radio3_1],
+    inputs=[textbox, "microphone"],
     outputs=["audio"],
 )