Spaces:
Running
on
Zero
Running
on
Zero
raoyonghui
commited on
Commit
•
dc256a6
1
Parent(s):
36f9ba6
Update app.py
Browse files
app.py
CHANGED
@@ -28,11 +28,21 @@ device = torch.device("cuda" if torch.cuda.is_available() else "CPU")
|
|
28 |
|
29 |
whisper_model = whisper.load_model("turbo")
|
30 |
|
31 |
-
def detect_speech_language(
|
32 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
|
34 |
def detect_text_language(text):
|
35 |
-
langid.classify(text)[0]
|
36 |
|
37 |
@torch.no_grad()
|
38 |
def get_prompt_text(speech_16k, language):
|
@@ -41,7 +51,6 @@ def get_prompt_text(speech_16k, language):
|
|
41 |
short_prompt_end_ts = 0.0
|
42 |
|
43 |
asr_result = whisper_model.transcribe(speech_16k, language=language)
|
44 |
-
print("asr_result:", asr_result)
|
45 |
full_prompt_text = asr_result["text"] # whisper asr result
|
46 |
#text = asr_result["segments"][0]["text"] # whisperx asr result
|
47 |
shot_prompt_text = ""
|
@@ -51,8 +60,6 @@ def get_prompt_text(speech_16k, language):
|
|
51 |
short_prompt_end_ts = segment['end']
|
52 |
if short_prompt_end_ts >= 4:
|
53 |
break
|
54 |
-
print("full prompt text:", full_prompt_text, " shot_prompt_text:", shot_prompt_text,
|
55 |
-
"short_prompt_end_ts:", short_prompt_end_ts)
|
56 |
return full_prompt_text, shot_prompt_text, short_prompt_end_ts
|
57 |
|
58 |
|
@@ -310,7 +317,7 @@ def maskgct_inference(
|
|
310 |
speech_16k = librosa.load(prompt_speech_path, sr=16000)[0]
|
311 |
speech = librosa.load(prompt_speech_path, sr=24000)[0]
|
312 |
|
313 |
-
prompt_language = detect_speech_language(
|
314 |
full_prompt_text, short_prompt_text, shot_prompt_end_ts = get_prompt_text(prompt_speech_path,
|
315 |
prompt_language)
|
316 |
# use the first 4+ seconds wav as the prompt in case the prompt wav is too long
|
@@ -321,7 +328,7 @@ def maskgct_inference(
|
|
321 |
device,
|
322 |
speech_16k,
|
323 |
short_prompt_text,
|
324 |
-
|
325 |
target_text,
|
326 |
target_language,
|
327 |
target_len,
|
@@ -393,9 +400,17 @@ iface = gr.Interface(
|
|
393 |
outputs=gr.Audio(label="Generated Audio"),
|
394 |
title="MaskGCT TTS Demo",
|
395 |
description="""
|
396 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
397 |
"""
|
398 |
)
|
399 |
|
400 |
# Launch the interface
|
401 |
-
iface.launch(allowed_paths=["./output"])
|
|
|
28 |
|
29 |
whisper_model = whisper.load_model("turbo")
|
30 |
|
31 |
+
def detect_speech_language(speech_file):
|
32 |
+
# load audio and pad/trim it to fit 30 seconds
|
33 |
+
audio = whisper.load_audio(speech_file)
|
34 |
+
audio = whisper.pad_or_trim(audio)
|
35 |
+
|
36 |
+
# make log-Mel spectrogram and move to the same device as the model
|
37 |
+
mel = whisper.log_mel_spectrogram(audio, n_mels=128).to(whisper_model.device)
|
38 |
+
|
39 |
+
# detect the spoken language
|
40 |
+
_, probs = whisper_model.detect_language(mel)
|
41 |
+
return max(probs, key=probs.get)
|
42 |
+
|
43 |
|
44 |
def detect_text_language(text):
|
45 |
+
return langid.classify(text)[0]
|
46 |
|
47 |
@torch.no_grad()
|
48 |
def get_prompt_text(speech_16k, language):
|
|
|
51 |
short_prompt_end_ts = 0.0
|
52 |
|
53 |
asr_result = whisper_model.transcribe(speech_16k, language=language)
|
|
|
54 |
full_prompt_text = asr_result["text"] # whisper asr result
|
55 |
#text = asr_result["segments"][0]["text"] # whisperx asr result
|
56 |
shot_prompt_text = ""
|
|
|
60 |
short_prompt_end_ts = segment['end']
|
61 |
if short_prompt_end_ts >= 4:
|
62 |
break
|
|
|
|
|
63 |
return full_prompt_text, shot_prompt_text, short_prompt_end_ts
|
64 |
|
65 |
|
|
|
317 |
speech_16k = librosa.load(prompt_speech_path, sr=16000)[0]
|
318 |
speech = librosa.load(prompt_speech_path, sr=24000)[0]
|
319 |
|
320 |
+
prompt_language = detect_speech_language(prompt_speech_path)
|
321 |
full_prompt_text, short_prompt_text, shot_prompt_end_ts = get_prompt_text(prompt_speech_path,
|
322 |
prompt_language)
|
323 |
# use the first 4+ seconds wav as the prompt in case the prompt wav is too long
|
|
|
328 |
device,
|
329 |
speech_16k,
|
330 |
short_prompt_text,
|
331 |
+
prompt_language,
|
332 |
target_text,
|
333 |
target_language,
|
334 |
target_len,
|
|
|
400 |
outputs=gr.Audio(label="Generated Audio"),
|
401 |
title="MaskGCT TTS Demo",
|
402 |
description="""
|
403 |
+
## MaskGCT: Zero-Shot Text-to-Speech with Masked Generative Codec Transformer
|
404 |
+
|
405 |
+
[![arXiv](https://img.shields.io/badge/arXiv-Paper-COLOR.svg)](https://arxiv.org/abs/2409.00750)
|
406 |
+
|
407 |
+
[![hf](https://img.shields.io/badge/%F0%9F%A4%97%20HuggingFace-model-yellow)](https://huggingface.co/amphion/maskgct)
|
408 |
+
|
409 |
+
[![hf](https://img.shields.io/badge/%F0%9F%A4%97%20HuggingFace-demo-pink)](https://huggingface.co/spaces/amphion/maskgct)
|
410 |
+
|
411 |
+
[![readme](https://img.shields.io/badge/README-Key%20Features-blue)](https://github.com/open-mmlab/Amphion/tree/main/models/tts/maskgct)
|
412 |
"""
|
413 |
)
|
414 |
|
415 |
# Launch the interface
|
416 |
+
iface.launch(allowed_paths=["./output"])
|