CNN

app.py

@@ -1,5 +1,6 @@
 import gradio as gr
 import argparse
+import torchaudio
 import json
 import datetime as dt
 import numpy as np
@@ -67,7 +68,7 @@ spekears = ['Madi', 'Marzhan', 'Akzhol']
 
 def generate_audio(text, quantity, speaker, emotion_1, emotion_2):
     x, x_lengths = convert_text(text)
-    emo_1, emo_2 = emotion_1, emotion_2
+    emo_1, emo_2 = emotions.index(emotion_1), emotions.index(emotion_2)
     emo1 = torch.LongTensor([emo_1]).to(device)
     emo2 = torch.LongTensor([emo_2]).to(device)
     sid = torch.LongTensor([spekears.index(speaker)]).to(device)
@@ -97,6 +98,13 @@ def generate_audio(text, quantity, speaker, emotion_1, emotion_2):
     sr = 22050
     return (sr, audio)
 
+
+
+def audio_check():
+    wav_path = '/Users/adalabilbekov/Desktop/ISSAI/experiments_yet/neutral_happy_time10_guid10_temp20.wav'
+    waveform, sr = torchaudio.load(wav_path)
+    return (sr, np.flipud(waveform))
+
 # def sentence_builder(quantity, emotion_1, emotion_2):
 #     return f"""The {quantity} {emotion_1}s from {" and ".join(emotion_2)}"""
 
@@ -113,5 +121,5 @@ demo = gr.Interface(
     ],
     "audio",
 )
-
+print('launching the app')
 demo.launch()

configs/train_grad.json

@@ -63,6 +63,6 @@
     "h_decoder": 4,
     "decoder_dropout":0.1,
 
-    "classifier_type": "CNN-with-time"
+    "classifier_type": "CNN"
   }
 }