brige/human
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases 2 Wiki Activity

添加测试的代码

Browse Source
This commit is contained in:
brige 2024-09-30 01:45:49 +08:00
parent 3aec7b7103
commit a68fbbc0de
1 changed files with 16 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
Human.py
View File

@ -254,8 +254,8 @@ def datagen_signal(frame, mel, face_det_results):
face, coord = face_det_results[idx].copy()
face = cv2.resize(face, (img_size, img_size))
m = mel
for i, m in enumerate(mel):
img_batch.append(face)
mel_batch.append(m)
frame_batch.append(frame_to_save)
@ -342,6 +342,7 @@ class Human:
def inter(self, model, chunks, face_list_cycle, face_det_results, out, j):
inputs = np.concatenate(chunks) # [5 * chunk]
mel = audio.melspectrogram(inputs)
print("inter", len(mel[0]))
if np.isnan(mel.reshape(-1)).sum() > 0:
raise ValueError(
'Mel contains nan! Using a TTS voice? Add a small epsilon noise to the wav file and try again')
@ -352,19 +353,21 @@ class Human:
mel_idx_multiplier = 80. / self._fps
print('mel_idx_multiplier:', mel_idx_multiplier)
i = 0
mel_chunks = []
while 1:
start_idx = int(i * mel_idx_multiplier)
if start_idx + mel_step_size > len(mel[0]):
# mel_chunks.append(mel[:, len(mel[0]) - mel_step_size:])
self.mel_chunks_queue_.put(mel[:, len(mel[0]) - mel_step_size:])
mel_chunks.append(mel[:, len(mel[0]) - mel_step_size:])
# self.mel_chunks_queue_.put(mel[:, len(mel[0]) - mel_step_size:])
break
# mel_chunks.append(mel[:, start_idx: start_idx + mel_step_size])
self.mel_chunks_queue_.put(mel[:, start_idx: start_idx + mel_step_size])
mel_chunks.append(mel[:, start_idx: start_idx + mel_step_size])
# self.mel_chunks_queue_.put(mel[:, start_idx: start_idx + mel_step_size])
i += 1
self.mel_chunks_queue_.put(mel_chunks)
while not self.mel_chunks_queue_.empty():
print("self.mel_chunks_queue_ len:", self.mel_chunks_queue_.qsize())
m = self.mel_chunks_queue_.get()
# mel_batch = np.reshape(m, [len(m), mel_batch.shape[1], mel_batch.shape[2], 1])
img_batch, mel_batch, frames, coords = datagen_signal(face_list_cycle[0], m, face_det_results)
img_batch = torch.FloatTensor(np.transpose(img_batch, (0, 3, 1, 2))).to(device)
@ -413,7 +416,8 @@ class Human:
print('wav length:', stream_len)
_audio_chunk_queue = queue.Queue()
index = 0
chunk_len = 6400
chunk_len = 640# // 200
print('chunk_len:', chunk_len)
while stream_len >= chunk_len:
audio_chunk = stream[index:index + chunk_len]
_audio_chunk_queue.put(audio_chunk)
@ -430,7 +434,7 @@ class Human:
j = 0
while not _audio_chunk_queue.empty():
chunks = []
length = min(5, _audio_chunk_queue.qsize())
length = min(64, _audio_chunk_queue.qsize())
for i in range(length):
chunks.append(_audio_chunk_queue.get())