File: rtp_audio.py

package info (click to toggle)
python-voip-utils 0.3.4-1
  • links: PTS, VCS
  • area: main
  • in suites: forky, sid
  • size: 640 kB
  • sloc: python: 1,976; makefile: 2
file content (252 lines) | stat: -rw-r--r-- 7,475 bytes parent folder | download | duplicates (2)
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
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
"""Utility for converting audio to/from RTP + OPUS packets."""

try:
    # Use built-in audioop until it's removed in Python 3.13
    import audioop  # pylint: disable=deprecated-module
except ImportError:
    from . import pyaudioop as audioop  # type: ignore[no-redef]

import logging
import random
import struct
from collections.abc import Iterable
from dataclasses import dataclass
from typing import Any

import opuslib

from .const import OPUS_PAYLOAD_TYPE
from .error import RtpError

_LOGGER = logging.getLogger(__name__)


@dataclass
class RtpOpusInput:
    """Extracts audio from RTP packets with OPUS."""

    opus_rate: int = 48000  # Hz
    opus_width: int = 2  # bytes
    opus_channels: int = 2
    opus_frame_size: int = 960  # samples per channel
    opus_payload_type: int = OPUS_PAYLOAD_TYPE  # set by GrandStream

    def __post_init__(
        self,
    ) -> None:
        """Initialize encoder and state."""
        self._decoder = opuslib.api.decoder.create_state(
            self.opus_rate, self.opus_channels
        )

    def process_packet(
        self,
        rtp_bytes: bytes,
        rate: int,
        width: int,
        channels: int,
    ) -> bytes:
        """Extract, decode, and return raw audio from RTP packet."""
        if channels not in (1, 2):
            raise ValueError("Only mono and stereo audio is supported")

        # Minimum header size
        if len(rtp_bytes) < 12:
            raise RtpError("RTP packet is too small")

        # See: https://en.wikipedia.org/wiki/Real-time_Transport_Protocol#Packet_header
        flags, payload_type, _sequence_num, _timestamp, _ssrc = struct.unpack(
            ">BBHLL", rtp_bytes[:12]
        )

        if flags != 0b10000000:
            raise RtpError("Padding and extension headers not supported")

        payload_type &= 0x7F  # Remove marker bit
        if payload_type != self.opus_payload_type:
            raise RtpError(
                f"Expected payload type {self.opus_payload_type}, got {payload_type}"
            )

        # Assume no padding, extension headers, etc.
        opus_bytes = rtp_bytes[12:]

        # Decode into raw audio.
        # This will always be 48Khz stereo with 16-bit samples.
        audio_bytes = opuslib.api.decoder.decode(
            self._decoder,
            opus_bytes,
            len(opus_bytes),
            self.opus_frame_size,
            False,  # no forward error correction (fec)
        )

        # Convert to target sample rate, etc.
        if channels == 1:
            # Convert to mono
            audio_bytes = audioop.tomono(
                audio_bytes,
                self.opus_width,
                1.0,
                1.0,
            )

        if rate != self.opus_rate:
            # Resample
            audio_bytes, _state = audioop.ratecv(
                audio_bytes,
                self.opus_width,
                channels,
                self.opus_rate,
                rate,
                None,
            )

        if width != self.opus_width:
            # Resize
            audio_bytes = audioop.lin2lin(
                audio_bytes,
                self.opus_width,
                width,
            )

        return audio_bytes


@dataclass
class RtpOpusOutput:
    """Prepares audio to send to an RTP client using OPUS."""

    opus_rate: int = 48000  # Hz
    opus_width: int = 2  # bytes
    opus_channels: int = 2
    opus_frame_size: int = 960  # samples per channel
    opus_payload_type: int = OPUS_PAYLOAD_TYPE  # set by GrandStream
    opus_bytes_per_frame: int = 960 * 2 * 2  # 16-bit x stereo

    _rtp_flags: int = 0b10000000  # v2, no padding/extensions/CSRCs
    _rtp_sequence_num: int = 0
    _rtp_timestamp: int = 0
    _rtp_ssrc: int = 0

    _encoder: opuslib.api.encoder.Encoder = None
    _audio_buffer: bytes = None  # type: ignore[assignment]
    _resample_state: Any = None

    def __post_init__(
        self,
    ) -> None:
        """Initialize encoder and state."""
        self.opus_bytes_per_frame = (
            self.opus_frame_size * self.opus_width * self.opus_channels
        )

        # Set up OPUS encoder for VoIP
        self._encoder = opuslib.api.encoder.create_state(
            self.opus_rate,
            self.opus_width,
            opuslib.APPLICATION_VOIP,
        )

        self.reset()

    def reset(self):
        """Clear audio buffer and state."""
        self._audio_buffer = b""
        self._resample_state = None

        # Recommended to start from random offsets to aid encryption
        self._rtp_sequence_num = random.randint(0, 2**10)
        self._rtp_timestamp = random.randint(1, 2**10)

        # Change each time
        self._rtp_ssrc = random.randint(0, 2**32)

    def process_audio(
        self,
        audio_bytes: bytes,
        rate: int,
        width: int,
        channels: int,
        is_end: bool = False,
    ) -> Iterable[bytes]:
        """Process a chunk of raw audio and yield RTP packet(s)."""
        if rate != self.opus_rate:
            # Convert to 48Khz
            audio_bytes, self._resample_state = audioop.ratecv(
                audio_bytes,
                width,
                channels,
                rate,
                self.opus_rate,
                self._resample_state,
            )

        if width != self.opus_width:
            # Adjust sample width
            audio_bytes = audioop.lin2lin(
                audio_bytes,
                width,
                self.opus_width,
            )

        if channels != self.opus_channels:
            # Convert to stereo
            audio_bytes = audioop.tostereo(
                audio_bytes,
                self.opus_width,
                1.0,
                1.0,
            )

        self._audio_buffer += audio_bytes
        if is_end:
            # Pad with silence
            bytes_missing = len(self._audio_buffer) % self.opus_bytes_per_frame
            if bytes_missing > 0:
                self._audio_buffer += bytes(bytes_missing)

        num_frames = len(self._audio_buffer) // self.opus_bytes_per_frame

        # Process chunks with *exactly* the desired number of frames
        for i in range(num_frames):
            offset = i * self.opus_bytes_per_frame
            audio_chunk = self._audio_buffer[
                offset : offset + self.opus_bytes_per_frame
            ]

            # Encode to OPUS packet
            opus_bytes = opuslib.api.encoder.encode(
                self._encoder,
                audio_chunk,
                self.opus_frame_size,
                4000,  # recommended in opus docs
            )

            # Add RTP header
            # See: https://en.wikipedia.org/wiki/Real-time_Transport_Protocol#Packet_header
            rtp_bytes = struct.pack(
                ">BBHLL",
                self._rtp_flags,
                self.opus_payload_type,
                self._rtp_sequence_num,
                self._rtp_timestamp,
                self._rtp_ssrc,
            )

            # RTP packet
            yield rtp_bytes + opus_bytes

            # Next frame
            self._rtp_sequence_num += 1
            self._rtp_timestamp += self.opus_frame_size

        if num_frames > 0:
            # Remove audio already sent
            self._audio_buffer = self._audio_buffer[
                num_frames * self.opus_bytes_per_frame :
            ]

        if is_end:
            # Clear audio buffer and state
            self.reset()