updates

app.py

@@ -1,6 +1,7 @@
 import os
 import torch
 import gradio as gr
+import numpy as np
 import os.path as op
 import pyarabic.araby as araby
 
@@ -18,6 +19,8 @@ checkpoint['cfg']['task'].t5_task = 't2s'
 checkpoint['cfg']['task'].hubert_label_dir = "utils/"
 checkpoint['cfg']['task'].bpe_tokenizer = "utils/arabic.model"
 checkpoint['cfg']['task'].data = "utils/"
+checkpoint['cfg']['model'].mask_prob = 0.0
+checkpoint['cfg']['task'].mask_prob = 0.0
 task = ArTSTTask.setup_task(checkpoint['cfg']['task'])
 
 emb_path='embs/clartts.npy'
@@ -43,6 +46,8 @@ def process_text(text):
 net_input = {}
 
 def inference(text, spkr=emb_path):
+    if len(text.strip()) == 0:
+        return (16000, np.zeros(0).astype(np.int16))
     net_input['src_tokens'] = process_text(text)
     net_input['spkembs'] = get_embs(spkr)
     outs, _, attn = task.generate_speech(

requirements.txt

@@ -30,4 +30,5 @@ nvidia-nccl-cu11==2.14.3
 nvidia-nvtx-cu11==11.7.91
 tensorboardx==2.6
 transformers
-speechbrain
+speechbrain
+numpy==1.23.5