Spaces:
Running
on
Zero
Running
on
Zero
Commit
·
f447f4e
1
Parent(s):
d2c9151
code update for duration of ZeroGPU
Browse files- tts/gradio_api.py +3 -4
- tts/infer_cli.py +6 -0
tts/gradio_api.py
CHANGED
@@ -20,9 +20,7 @@ import gradio as gr
|
|
20 |
import traceback
|
21 |
from tts.infer_cli import MegaTTS3DiTInfer, convert_to_wav, cut_wav
|
22 |
|
23 |
-
import spaces
|
24 |
|
25 |
-
@spaces.GPU(duration=120)
|
26 |
def model_worker(input_queue, output_queue, device_id):
|
27 |
device = None
|
28 |
if device_id is not None:
|
@@ -39,8 +37,9 @@ def model_worker(input_queue, output_queue, device_id):
|
|
39 |
cut_wav(wav_path, max_len=28)
|
40 |
with open(wav_path, 'rb') as file:
|
41 |
file_content = file.read()
|
42 |
-
|
43 |
-
|
|
|
44 |
output_queue.put(wav_bytes)
|
45 |
except Exception as e:
|
46 |
traceback.print_exc()
|
|
|
20 |
import traceback
|
21 |
from tts.infer_cli import MegaTTS3DiTInfer, convert_to_wav, cut_wav
|
22 |
|
|
|
23 |
|
|
|
24 |
def model_worker(input_queue, output_queue, device_id):
|
25 |
device = None
|
26 |
if device_id is not None:
|
|
|
37 |
cut_wav(wav_path, max_len=28)
|
38 |
with open(wav_path, 'rb') as file:
|
39 |
file_content = file.read()
|
40 |
+
wav_bytes = infer_pipe.forward_zerogpu(file_content, inp_npy_path, inp_text, time_step=infer_timestep, p_w=p_w, t_w=t_w)
|
41 |
+
# resource_context = infer_pipe.preprocess(file_content, latent_file=inp_npy_path)
|
42 |
+
# wav_bytes = infer_pipe.forward(resource_context, inp_text, time_step=infer_timestep, p_w=p_w, t_w=t_w)
|
43 |
output_queue.put(wav_bytes)
|
44 |
except Exception as e:
|
45 |
traceback.print_exc()
|
tts/infer_cli.py
CHANGED
@@ -252,6 +252,12 @@ class MegaTTS3DiTInfer():
|
|
252 |
wav_pred = combine_audio_segments(wav_pred_, sr=self.sr).astype(float)
|
253 |
return to_wav_bytes(wav_pred, self.sr)
|
254 |
|
|
|
|
|
|
|
|
|
|
|
|
|
255 |
|
256 |
if __name__ == '__main__':
|
257 |
parser = argparse.ArgumentParser()
|
|
|
252 |
wav_pred = combine_audio_segments(wav_pred_, sr=self.sr).astype(float)
|
253 |
return to_wav_bytes(wav_pred, self.sr)
|
254 |
|
255 |
+
@spaces.GPU(duration=120)
|
256 |
+
def forward_zerogpu(self, file_content, latent_file, inp_text, time_step, p_w, t_w):
|
257 |
+
resource_context = self.preprocess(file_content, latent_file)
|
258 |
+
wav_bytes = self.forward(resource_context, inp_text, time_step=time_step, p_w=p_w, t_w=t_w)
|
259 |
+
return wav_bytes
|
260 |
+
|
261 |
|
262 |
if __name__ == '__main__':
|
263 |
parser = argparse.ArgumentParser()
|