1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
|
import cython
from cython.cimports.av.audio.format import get_audio_format
from cython.cimports.av.audio.layout import get_audio_layout
from cython.cimports.av.audio.plane import AudioPlane
from cython.cimports.av.error import err_check
from cython.cimports.av.utils import check_ndarray
_cinit_bypass_sentinel = object()
@cython.cfunc
def alloc_audio_frame() -> AudioFrame:
return AudioFrame(_cinit_bypass_sentinel)
format_dtypes = {
"dbl": "f8",
"dblp": "f8",
"flt": "f4",
"fltp": "f4",
"s16": "i2",
"s16p": "i2",
"s32": "i4",
"s32p": "i4",
"u8": "u1",
"u8p": "u1",
}
@cython.cclass
class AudioFrame(Frame):
"""A frame of audio."""
def __cinit__(self, format="s16", layout="stereo", samples=0, align=1):
if format is _cinit_bypass_sentinel:
return
cy_format: AudioFormat = AudioFormat(format)
cy_layout: AudioLayout = AudioLayout(layout)
self._init(cy_format.sample_fmt, cy_layout.layout, samples, align)
@cython.cfunc
def _init(
self,
format: lib.AVSampleFormat,
layout: lib.AVChannelLayout,
nb_samples: cython.uint,
align: cython.uint,
):
self.ptr.nb_samples = nb_samples
self.ptr.format = format
self.ptr.ch_layout = layout
# Sometimes this is called twice. Oh well.
self._init_user_attributes()
if self.layout.nb_channels != 0 and nb_samples:
# Cleanup the old buffer.
lib.av_freep(cython.address(self._buffer))
# Get a new one.
self._buffer_size = err_check(
lib.av_samples_get_buffer_size(
cython.NULL, self.layout.nb_channels, nb_samples, format, align
)
)
self._buffer = cython.cast(
cython.pointer[uint8_t], lib.av_malloc(self._buffer_size)
)
if not self._buffer:
raise MemoryError("cannot allocate AudioFrame buffer")
# Connect the data pointers to the buffer.
err_check(
lib.avcodec_fill_audio_frame(
self.ptr,
self.layout.nb_channels,
cython.cast(lib.AVSampleFormat, self.ptr.format),
self._buffer,
self._buffer_size,
align,
)
)
def __dealloc__(self):
lib.av_freep(cython.address(self._buffer))
@cython.cfunc
def _init_user_attributes(self):
self.layout = get_audio_layout(self.ptr.ch_layout)
self.format = get_audio_format(cython.cast(lib.AVSampleFormat, self.ptr.format))
def __repr__(self):
return (
f"<av.{self.__class__.__name__} pts={self.pts}, {self.samples} "
f"samples at {self.rate}Hz, {self.layout.name}, {self.format.name} at 0x{id(self):x}>"
)
@staticmethod
def from_ndarray(array, format="s16", layout="stereo"):
"""
Construct a frame from a numpy array.
"""
import numpy as np
py_format = format if isinstance(format, AudioFormat) else AudioFormat(format)
py_layout = layout if isinstance(layout, AudioLayout) else AudioLayout(layout)
format = py_format.name
# map avcodec type to numpy type
try:
dtype = np.dtype(format_dtypes[format])
except KeyError:
raise ValueError(
f"Conversion from numpy array with format `{format}` is not yet supported"
)
# check input format
nb_channels = py_layout.nb_channels
check_ndarray(array, dtype, 2)
if py_format.is_planar:
if array.shape[0] != nb_channels:
raise ValueError(
f"Expected planar `array.shape[0]` to equal `{nb_channels}` but got `{array.shape[0]}`"
)
samples = array.shape[1]
else:
if array.shape[0] != 1:
raise ValueError(
f"Expected packed `array.shape[0]` to equal `1` but got `{array.shape[0]}`"
)
samples = array.shape[1] // nb_channels
frame = AudioFrame(format=py_format, layout=py_layout, samples=samples)
for i, plane in enumerate(frame.planes):
plane.update(array[i, :])
return frame
@property
def planes(self):
"""
A tuple of :class:`~av.audio.plane.AudioPlane`.
:type: tuple
"""
plane_count: cython.int = 0
while self.ptr.extended_data[plane_count]:
plane_count += 1
return tuple([AudioPlane(self, i) for i in range(plane_count)])
@property
def samples(self):
"""
Number of audio samples (per channel).
:type: int
"""
return self.ptr.nb_samples
@property
def sample_rate(self):
"""
Sample rate of the audio data, in samples per second.
:type: int
"""
return self.ptr.sample_rate
@sample_rate.setter
def sample_rate(self, value):
self.ptr.sample_rate = value
@property
def rate(self):
"""Another name for :attr:`sample_rate`."""
return self.ptr.sample_rate
@rate.setter
def rate(self, value):
self.ptr.sample_rate = value
def to_ndarray(self):
"""Get a numpy array of this frame.
.. note:: Numpy must be installed.
"""
import numpy as np
try:
dtype = np.dtype(format_dtypes[self.format.name])
except KeyError:
raise ValueError(
f"Conversion from {self.format.name!r} format to numpy array is not supported."
)
if self.format.is_planar:
count = self.samples
else:
count = self.samples * self.layout.nb_channels
return np.vstack(
[np.frombuffer(x, dtype=dtype, count=count) for x in self.planes]
)
|
