Update app.py
Browse files
app.py
CHANGED
|
@@ -1,7 +1,28 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
|
| 3 |
-
|
| 4 |
-
|
|
|
|
| 5 |
|
| 6 |
-
|
| 7 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
|
| 3 |
+
# Load spectrogram generator
|
| 4 |
+
from nemo.collections.tts.models import FastPitchModel
|
| 5 |
+
spec_generator = FastPitchModel.restore_from("inOXcrm/German_multispeaker_FastPitch_nemo")
|
| 6 |
|
| 7 |
+
# Load Vocoder
|
| 8 |
+
from nemo.collections.tts.models import HifiGanModel
|
| 9 |
+
model = HifiGanModel.from_pretrained(model_name="tts_de_hui_hifigan_ft_fastpitch_multispeaker_5")
|
| 10 |
+
|
| 11 |
+
# Generate audio
|
| 12 |
+
|
| 13 |
+
def generate_audio(speaker_id, input_txt):
|
| 14 |
+
sr=44100
|
| 15 |
+
parsed = spec_generator.parse(input_txt)
|
| 16 |
+
spectrogram = spec_generator.generate_spectrogram(tokens=parsed, speaker=speaker_id)
|
| 17 |
+
audio = model.convert_spectrogram_to_audio(spec=spectrogram)
|
| 18 |
+
return (sr, audio)
|
| 19 |
+
|
| 20 |
+
|
| 21 |
+
gr.Interface(
|
| 22 |
+
generate_audio,
|
| 23 |
+
[
|
| 24 |
+
gr.Textbox(type="text", value=1, label="Speaker ID (0-5)")
|
| 25 |
+
gr.Textbox(type="text", value=1, label="Input Text"),
|
| 26 |
+
],
|
| 27 |
+
"audio",
|
| 28 |
+
).launch()
|