numpy audio

app.py

@@ -7,6 +7,11 @@ from datasets import load_dataset
 import soundfile as sf
 import torch
 import os
+import io 
+import base64
+import numpy as np
+from pydub import AudioSegment
+
 os.environ['TRANSFORMERS_CACHE'] = '.cache'
 
 print ("----- setting up pipeline -----")
@@ -30,17 +35,44 @@ print ("----- synthetizing audio -----")
 def greet(name):
     return "Hello " + name + "!!"
 
+
 def synthesise_audio(text, forward_params=None):
+    if len(text) > 100:
+        raise ValueError("Error: El texto es demasiado largo. Por favor, limita tu entrada a 100 caracteres.")
+
     speech = synthesiser(text, forward_params={"speaker_embeddings": speaker_embedding})
-    sf.write("speech.wav", speech["audio"], samplerate=speech["sampling_rate"])
-    return "speech.wav"
+    # sf.write("speech.wav", speech["audio"], samplerate=speech["sampling_rate"])
+    # return "speech.wav"
+    
+    # Convert numpy array to audio
+    #with io.BytesIO() as f:
+    #    sf.write(f, speech["audio"], samplerate=speech["sampling_rate"], format='wav')
+    #    audio = f.getvalue()
+
+    # Convert numpy array to audio
+    audio = np.int16(speech["audio"] * 32767).tobytes()
+    audio_segment = AudioSegment(audio, sample_width=2, frame_rate=speech["sampling_rate"], channels=1)
+
+
+
+    #return speech["audio"]
+
+    return audio
+    
+  
+    
+
 
 #demo = gr.Interface(fn=greet, inputs="text", outputs="text", description="----- TTS Testing -----")
 
+input_text = gr.Textbox(lines=10, label="Enter text here")
+
+
 demo = gr.Interface(fn=synthesise_audio,
-                    inputs="text",
-                    outputs="audio",
+                    inputs=input_text,
+                    #outputs="audio",
+                    outputs = gr.Audio(type="numpy"),
                     description="----- manuai Text To Speech generator -----",
                     allow_flagging = False)
 
-demo.launch()
+demo.launch(debug = True)