Update app.py

app.py

@@ -4,9 +4,9 @@ import os
 from groq import Groq
 import numpy as np
 import wave
+import uuid
 
 #tts
-import tempfile
 import torchaudio
 #from speechbrain.inference.TTS import FastSpeech2
 from speechbrain.inference.TTS import Tacotron2
@@ -17,19 +17,13 @@ tacotron2 = Tacotron2.from_hparams(source="speechbrain/tts-tacotron2-ljspeech",
 hifi_gan = HIFIGAN.from_hparams(source="speechbrain/tts-hifigan-ljspeech", savedir="pretrained_models/tts-hifigan-ljspeech")
 
 def text_to_speech(text):
-    with tempfile.NamedTemporaryFile(delete=False, suffix='.wav') as temp_file:
-        # mel_output, durations, pitch, energy = fastspeech2.encode_text(
-        #   [text],
-        #   pace=1.0,        # scale up/down the speed
-        #   pitch_rate=1.0,  # scale up/down the pitch
-        #   energy_rate=1.0, # scale up/down the energy
-        # )
-        mel_output, mel_length, alignment = tacotron2.encode_text(text)
-        # Running Vocoder (spectrogram-to-waveform)
-        waveforms = hifi_gan.decode_batch(mel_output)
-        # Save the waverform
-        torchaudio.save(temp_file.name, waveforms.squeeze(1), 22050)
-    return temp_file.name
+    mel_output, mel_length, alignment = tacotron2.encode_text(text)
+    # Running Vocoder (spectrogram-to-waveform)
+    waveforms = hifi_gan.decode_batch(mel_output)
+    # Save the waverform
+    outfile = f"{uuid.uuid4()}.wav"
+    torchaudio.save(outfile, waveforms.squeeze(1), 22050)
+    return outfile
 
 def combine_audio_files(audio_files):
     data= []