Spaces:
Paused
Paused
myhanhhyugen
commited on
Commit
•
fa0ce57
1
Parent(s):
66521e9
create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from TTSInferencing import TTSInferencing
|
2 |
+
import streamlit as st
|
3 |
+
from speechbrain.inference.vocoders import HIFIGAN
|
4 |
+
import torchaudio
|
5 |
+
|
6 |
+
|
7 |
+
tts_model = TTSInferencing.from_hparams(source="./",
|
8 |
+
hparams_file='./hyperparams.yaml',
|
9 |
+
pymodule_file='./module_classes.py',
|
10 |
+
savedir="/results/",
|
11 |
+
)
|
12 |
+
|
13 |
+
hifi_gan = HIFIGAN.from_hparams(source="speechbrain/tts-hifigan-ljspeech", savedir='/pretrained_models/hifi-gan-ljspeech')
|
14 |
+
|
15 |
+
# text = ["Hello I am a girl", "How is your day going", "I hope you are doing well"]
|
16 |
+
|
17 |
+
# Input text
|
18 |
+
text = [st.text_input("Enter your text here")]
|
19 |
+
|
20 |
+
if text:
|
21 |
+
mel_outputs = tts_model.encode_batch(text)
|
22 |
+
waveforms = hifi_gan.decode_batch(mel_outputs)
|
23 |
+
|
24 |
+
waveform = waveforms[0].squeeze(1)
|
25 |
+
|
26 |
+
# Display the audio widget to play the synthesized speech
|
27 |
+
st.audio(waveform.numpy(), format="audio/wav")
|