from malaya_speech.model.frame import Frame
from herpetologist import check_type
from typing import List
[docs]@check_type
def frames(
audio,
frame_duration_ms: int = 30,
sample_rate: int = 16000,
append_ending_trail: bool = True,
):
"""
Generates audio frames from audio.
Takes the desired frame duration in milliseconds, the audio, and the sample rate.
Parameters
----------
audio: np.array
frame_duration_ms: int, optional (default=30)
sample_rate: int, optional (default=16000)
append_ending_trail: bool, optional (default=True)
if True, will append last trail and this last trail might not same length as `frame_duration_ms`.
Returns
-------
result: List[malaya_speech.model.frame.Frame]
"""
n = int(sample_rate * (frame_duration_ms / 1000.0))
offset = 0
timestamp = 0.0
duration = float(n) / sample_rate
results = []
while offset + n <= len(audio):
results.append(Frame(audio[offset: offset + n], timestamp, duration))
timestamp += duration
offset += n
if append_ending_trail and offset < len(audio):
results.append(
Frame(
audio[offset:], timestamp, len(audio) / sample_rate - timestamp
)
)
return results
[docs]def mel_sampling(
audio, frame_duration_ms=1200, overlap_ms=200, sample_rate=16000
):
"""
Generates audio frames from audio. This is for melspectrogram generative model.
Takes the desired frame duration in milliseconds, the audio, and the sample rate.
Parameters
----------
audio: np.array
frame_duration_ms: int, optional (default=1200)
overlap_ms: int, optional (default=200)
sample_rate: int, optional (default=16000)
Returns
-------
result: List[np.array]
"""
n = int(sample_rate * (frame_duration_ms / 1000.0))
n_overlap = int(sample_rate * (overlap_ms / 1000.0))
offset = 0
results = []
while offset + n <= len(audio):
results.append(audio[offset: offset + n])
offset += n - n_overlap
if offset < len(audio):
results.append(audio[offset:])
return results
[docs]def combine_mel_sampling(
samples, overlap_ms=200, sample_rate=16000, padded_ms=50
):
"""
To combine results from `mel_sampling`, output from melspectrogram generative model.
Parameters
----------
samples: List[np.array]
overlap_ms: int, optional (default=200)
sample_rate: int, optional (default=16000)
Returns
-------
result: List[np.array]
"""
n_overlap = int(sample_rate * (overlap_ms / 1000.0))
n_padded = int(sample_rate * (padded_ms / 1000.0))
results = []
for no, sample in enumerate(samples):
if no:
sample = sample[n_overlap - n_padded:]
results.append(sample[:-n_padded])
return results
