aboutsummaryrefslogtreecommitdiff
path: root/research/sample-data/mp3_psa_streamer.py
blob: 36d2974dbf6cd3fb943464ff3f8e08e02f162428 (plain)
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
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
#!/usr/bin/env python3
import os
import sys
import time
import socket
import struct
import argparse
import random
import tempfile
import subprocess
from tqdm import tqdm  # progress bar sexxy

# need that for lame transcoding
try:
    from mutagen.mp3 import MP3 # type: ignore
    MUTAGEN_AVAILABLE = True
except ImportError:
    MUTAGEN_AVAILABLE = False

# Default configuration values
DEFAULT_MULTICAST_ADDR = "172.16.20.109"
DEFAULT_PORT = 1681
DEFAULT_ZONE_INFO = "01 1000 0000 0000 0000 0000 0000"  # Zone 1
DEFAULT_SEND_DUPLICATES = True # Do it like Bodet SIGMA does and just send duplicates "for redundancy" (UDP for emergency systems who the fuck though that was a good idea)
DEFAULT_CHUNK_SIZE = 1000  # Need to use 1000 due to not understanding protocol fully yet
DEFAULT_TTL = 3 # We want it going to Multicast and then to the client, nomore otherwise bodeter speaker gets confused
DEFAULT_HEADER = "4d454c"  # MEL goofy ah header
DEFAULT_COMMAND = "0703"  # Probably the stream command
DEFAULT_STREAM_ID = "110c"  # Write none without qoutes to use random stream ID

# Prem prem maximum it supports
HIGH_QUALITY = {
    "bitrate": "256k",  # 256k max
    "sample_rate": 48000, # 48000 max
    "channels": 1,  # mono because uhm one speaker only
    "sample_format": "s32", # max s32
    "description": "Highest Quality (256kbps, 48kHz)"
}

# Bodeter Shitsy Sigma software defaults
LOW_QUALITY = {
    "bitrate": "64k", # Bodet Defaults
    "sample_rate": 32000,  # Bodet MP3 file default
    "channels": 1,
    "sample_format": "s16", # bodet MP3 file default
    "description": "Normal Quality (64kbps, 32kHz)"
}

DEFAULT_TIMING_FACTOR = 1  # Used back when chat gpt suggested i should adjust the timing... but it was really just that I have to use 1000 byte chunks ...

def compute_psa_checksum(data: bytes) -> bytes:
    """Compute PSA checksum for packet data."""
    var_e = 0x0000  # Starting seed value
    for i in range(len(data)):
        var_e ^= (data[i] + i) & 0xFFFF
    return var_e.to_bytes(2, 'big')  # 2-byte checksum, big-endian

def create_mp3_packet(sequence_num, zone_info, stream_id, mp3_chunk, show_packet_length=False):
    """Create a PSA packet containing MP3 data."""
    # Part 1: Header (static "MEL")
    header = bytes.fromhex("4d454c")
    
    # Prepare all the components that will go into the payload
    # Start marker (0100)
    start_marker = bytes.fromhex("0100")
    
    # Sequence number (1 byte, Little Endian with FF padding)
    seq = sequence_num.to_bytes(1, 'little') + b'\xff'
    
    # Command (static 070301)
    command = bytes.fromhex(DEFAULT_COMMAND)
    
    # Zone info
    zones = bytes.fromhex(zone_info)
    
    # Stream ID (2 bytes)
    stream_id_bytes = stream_id.to_bytes(2, 'little')
    
    # Constant data (appears in all original packets)
    constant_data = bytes.fromhex("05080503e8")
    
    
    # Build the payload (everything that comes after start_marker)
    payload = seq + command + zones + stream_id_bytes + constant_data + mp3_chunk
    
    # Calculate the length for the header - IMPORTANT: This matches what the PSA software does
    # Length is the total packet length MINUS the header (4d454c) and the length field itself (2 bytes)
    length_value = len(start_marker) + len(payload) + 7  # +2 for checksum
    
    # Insert the length as a 2-byte value (big endian)
    length_bytes = length_value.to_bytes(2, 'big')
    
    # Assemble the packet without checksum
    packet_data = header + length_bytes + start_marker + payload
    
    # Calculate and append checksum
    checksum = compute_psa_checksum(packet_data)
    
    # Create final packet
    final_packet = packet_data + checksum
    
    if show_packet_length:
        print(f"Packet length: {len(final_packet)} bytes, Length field: {length_value} (0x{length_value:04x})")
    
    return final_packet

def send_multicast_packet(sock, packet, multicast_addr, port, send_duplicates=True):
    """Send a packet to the multicast address."""
    try:
        # Send the packet
        sock.sendto(packet, (multicast_addr, port))
        
        # Send a duplicate (common in multicast to improve "reliability")
        if send_duplicates:
            sock.sendto(packet, (multicast_addr, port))
        
        return True
    except Exception as e:
        print(f"Error sending packet: {e}")
        return False

def setup_multicast_socket(ttl=DEFAULT_TTL):
    """Set up a socket for sending multicast UDP."""
    sock = socket.socket(socket.AF_INET, socket.SOCK_DGRAM, socket.IPPROTO_UDP)
    sock.setsockopt(socket.IPPROTO_IP, socket.IP_MULTICAST_TTL, ttl)
    return sock

def check_dependencies():
    """Check if required dependencies are installed."""
    try:
        subprocess.run(['lame', '--version'], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
    except FileNotFoundError:
        print("Error: LAME encoder is required but not found.")
        print("Please install LAME and make sure it's available in your PATH.")
        sys.exit(1)
    
    if not MUTAGEN_AVAILABLE:
        print("Warning: mutagen library not found. Limited audio file analysis available.")
        print("Install with: pip install mutagen")

def get_audio_info(mp3_file):
    """Get audio file information using mutagen if available."""
    try:
        if MUTAGEN_AVAILABLE:
            # Use mutagen to analyze the file
            audio = MP3(mp3_file)
            
            # Extract relevant information
            sample_rate = audio.info.sample_rate
            bit_rate = int(audio.info.bitrate / 1000)  # Convert to kbps
            channels = audio.info.channels
            codec_name = "mp3" if audio.info.layer == 3 else f"mpeg-{audio.info.layer}"
            
            return {
                'bit_rate': bit_rate,
                'sample_rate': sample_rate,
                'channels': channels,
                'codec_name': codec_name
            }
        else:
            # Fallback method: use LAME to identify file info
            cmd = ['lame', '--decode', mp3_file, '-t', '--brief', '-']
            result = subprocess.run(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.PIPE, text=True)
            
            # LAME outputs limited info to stderr when using --brief
            info_text = result.stderr
            
            # Very basic parsing - this is not ideal but works as a fallback
            codec_name = "mp3"  # Assume it's MP3
            sample_rate = 44100  # Default assumption
            channels = 2  # Default assumption
            bit_rate = 0  # Unknown
            
            for line in info_text.split('\n'):
                if "MPEG" in line and "Layer" in line:
                    codec_name = "mp3"
                if "Hz" in line:
                    parts = line.split()
                    for part in parts:
                        if "Hz" in part:
                            try:
                                sample_rate = int(part.replace("Hz", ""))
                            except ValueError:
                                pass
                if "stereo" in line.lower():
                    channels = 2
                elif "mono" in line.lower():
                    channels = 1
                if "kbps" in line:
                    parts = line.split()
                    for part in parts:
                        if "kbps" in part:
                            try:
                                bit_rate = int(part.replace("kbps", ""))
                            except ValueError:
                                pass
            
            return {
                'bit_rate': bit_rate,
                'sample_rate': sample_rate,
                'channels': channels,
                'codec_name': codec_name
            }
    
    except Exception as e:
        print(f"Error getting audio information: {e}")
        return None

def transcode_mp3(input_file, quality_preset, include_metadata=False):
    """Transcode MP3 file using LAME encoder."""
    print(f"Transcoding audio to {quality_preset['description']}...")
    
    # Create temporary output file
    fd, temp_output = tempfile.mkstemp(suffix='.mp3')
    os.close(fd)
    
    try:
        # Build LAME command
        cmd = [
            'lame',
            '--quiet',  # Less output
            '-m', 'm' if quality_preset['channels'] == 1 else 's',  # m=mono, s=stereo
            '--resample', str(quality_preset['sample_rate'] // 1000),  # Convert to kHz
            '-b', quality_preset['bitrate'].replace('k', ''),  # Remove 'k' suffix
            '--cbr',  # Use constant bitrate
        ]
        
        # Add options to minimize metadata if requested
        if not include_metadata:
            cmd.append('--noreplaygain')
            # Use more compatible tag options instead of --id3v2-none
            cmd.append('--tt')
            cmd.append('')  # Empty title
            cmd.append('--tc')
            cmd.append('')  # Empty comment
            cmd.append('--nohist')
        
        # Input and output files
        cmd.extend([input_file, temp_output])
        
        result = subprocess.run(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.PIPE, text=True)
        
        if result.returncode != 0:
            print(f"Error transcoding audio: {result.stderr}")
            os.remove(temp_output)
            return None
        
        return temp_output
    
    except Exception as e:
        print(f"Error during transcoding: {e}")
        if os.path.exists(temp_output):
            os.remove(temp_output)
        return None

def prepare_mp3_file(mp3_file, quality, include_metadata=False, sample_format=None):
    """Ensure the MP3 file meets the required specifications."""
    # Check if quality is valid
    quality_preset = HIGH_QUALITY if quality == "high" else LOW_QUALITY
    
    # Override sample format if specified
    if sample_format:
        quality_preset = quality_preset.copy()
        quality_preset["sample_format"] = sample_format
    
    # Get audio file information
    info = get_audio_info(mp3_file)
    if not info:
        return None
    
    print(f"Audio file details: {info['codec_name']}, {info['bit_rate']}kbps, "
          f"{info['sample_rate']}Hz, {info['channels']} channel(s)")
    
    # Check if transcoding is needed
    needs_transcode = False
    
    if info['codec_name'].lower() != 'mp3':
        print("Non-MP3 format detected. Transcoding required.")
        needs_transcode = True
    elif info['bit_rate'] is None or abs(info['bit_rate'] - int(quality_preset['bitrate'][:-1])) > 5:
        print(f"Bitrate mismatch: {info['bit_rate']}kbps vs {quality_preset['bitrate']}. Transcoding required.")
        needs_transcode = True
    elif info['sample_rate'] != quality_preset['sample_rate']:
        print(f"Sample rate mismatch: {info['sample_rate']}Hz vs {quality_preset['sample_rate']}Hz. Transcoding required.")
        needs_transcode = True
    elif info['channels'] != quality_preset['channels']:
        print(f"Channel count mismatch: {info['channels']} vs {quality_preset['channels']} (mono). Transcoding required.")
        needs_transcode = True
    
    # If transcoding is needed, do it
    if needs_transcode:
        return transcode_mp3(mp3_file, quality_preset, include_metadata)
    else:
        print("Audio file already meets required specifications.")
        return mp3_file

def calculate_delay_ms(chunk_size, bitrate, timing_factor=DEFAULT_TIMING_FACTOR):
    """
    Calculate the appropriate delay between chunks for real-time streaming.
    
    Args:
        chunk_size: Size of each audio chunk in bytes
        bitrate: Audio bitrate (string like '128k' or integer)
        timing_factor: Factor to adjust timing (lower = faster playback)
    
    Returns:
        Delay in milliseconds between packets
    """
    if isinstance(bitrate, str) and bitrate.endswith('k'):
        bitrate_kbps = int(bitrate[:-1])
    else:
        bitrate_kbps = int(bitrate)
    
    bytes_per_second = bitrate_kbps * 1000 // 8
    delay_ms = (chunk_size / bytes_per_second) * 1000 * timing_factor
    return delay_ms

class StreamTimer:
    """Maintains proper timing for streaming audio packets."""
    
    def __init__(self, chunk_size, bitrate, timing_factor=DEFAULT_TIMING_FACTOR):
        """Initialize the stream timer with audio parameters."""
        self.chunk_size = chunk_size
        self.bitrate = bitrate
        self.timing_factor = timing_factor
        self.start_time = None
        self.packets_sent = 0
        self.delay_per_packet = calculate_delay_ms(chunk_size, bitrate, timing_factor) / 1000.0
        
    def start(self):
        """Start the stream timer."""
        self.start_time = time.time()
        self.packets_sent = 0
        
    def wait_for_next_packet(self):
        """Wait until it's time to send the next packet."""
        if self.start_time is None:
            self.start()
            return
            
        self.packets_sent += 1
        
        # Calculate when this packet should be sent
        target_time = self.start_time + (self.packets_sent * self.delay_per_packet)
        
        # Calculate how long to wait
        now = time.time()
        wait_time = target_time - now
        
        # Only wait if we're ahead of schedule
        if wait_time > 0:
            time.sleep(wait_time)

def stream_mp3_to_psa(mp3_file, multicast_addr, port, zone_info, send_duplicates,
                      chunk_size, quality, include_metadata=False, sample_format=None,
                      show_packet_length=False, timing_factor=DEFAULT_TIMING_FACTOR, fixed_stream_id=None):
    """Stream an MP3 file to PSA system."""
    try:
        # Check dependencies
        check_dependencies()
        
        # Prepare MP3 file (transcode if needed)
        prepared_file = prepare_mp3_file(mp3_file, quality, include_metadata, sample_format)
        if not prepared_file:
            print("Failed to prepare MP3 file.")
            return
        
        using_temp_file = prepared_file != mp3_file
        
        try:
            # Use fixed stream ID if provided, otherwise generate a random one
            if fixed_stream_id is not None:
                stream_id = fixed_stream_id
            else:
                stream_id = random.randint(0, 0xFFFF)  # 16-bit (2 bytes) stream ID
            
            # Open the MP3 file
            with open(prepared_file, 'rb') as f:
                mp3_data = f.read()
            
            # Calculate number of chunks
            chunks = chunk_mp3_data(mp3_data, chunk_size)
            total_chunks = len(chunks)
            
            # Get the appropriate quality preset
            quality_preset = HIGH_QUALITY if quality == "high" else LOW_QUALITY
            
            # Calculate the appropriate delay for real-time streaming using the actual bitrate
            delay_ms = calculate_delay_ms(chunk_size, quality_preset['bitrate'])
            
            print(f"Streaming {os.path.basename(mp3_file)} ({len(mp3_data)/1024:.2f} KB)")
            print(f"Split into {total_chunks} packets of {chunk_size} bytes each")
            print(f"Sending to multicast {multicast_addr}:{port}")
            print(f"Stream ID: {stream_id:04x}, Zone info: {zone_info}")
            print(f"Duplicate packets: {'Yes' if send_duplicates else 'No'}")
            print(f"Quality preset: {HIGH_QUALITY['description'] if quality == 'high' else LOW_QUALITY['description']}")
            print(f"Including metadata: {'Yes' if include_metadata else 'No'}")
            print(f"Real-time streaming delay: {delay_ms:.2f}ms per packet")
            
            # Setup multicast socket
            sock = setup_multicast_socket()
            
            # Initialize the stream timer with the appropriate parameters
            timer = StreamTimer(chunk_size, quality_preset['bitrate'], timing_factor)
            timer.start()
            
            # Process and send each chunk
            sequence_num = 0
            with tqdm(total=total_chunks, desc="Streaming progress") as pbar:
                try:
                    for i, chunk in enumerate(chunks):
                        # Create packet with current sequence number
                        packet = create_mp3_packet(sequence_num, zone_info, stream_id, chunk, show_packet_length)
                        
                        # Send the packet
                        success = send_multicast_packet(sock, packet, multicast_addr, port, send_duplicates)
                        
                        # Increment sequence num
                        sequence_num = (sequence_num + 1) % 256
                        
                        # Update progress bar
                        pbar.update(1)
                        
                        # Wait for the next packet timing using our timer
                        if i < total_chunks - 1:  # No need to wait after the last chunk
                            timer.wait_for_next_packet()
                except Exception as e:
                    print(f"Error during streaming: {e}")
            
            print("\nStreaming completed successfully!")
            
        finally:
            # Clean up temporary file if created
            if using_temp_file and os.path.exists(prepared_file):
                os.remove(prepared_file)
                
    except FileNotFoundError:
        print(f"Error: MP3 file {mp3_file} not found")
        sys.exit(1)
    except Exception as e:
        print(f"Error during streaming: {e}")
        sys.exit(1)

def chunk_mp3_data(data, chunk_size):
    """
    Split MP3 data into chunks of specified size.
    
    Args:
        data: The complete MP3 data as bytes
        chunk_size: Size of each chunk in bytes
    
    Returns:
        A list of data chunks
    """
    chunks = []
    # Calculate how many chunks we'll have
    num_chunks = (len(data) + chunk_size - 1) // chunk_size
    
    for i in range(num_chunks):
        start = i * chunk_size
        end = min((i + 1) * chunk_size, len(data))
        chunks.append(data[start:end])
    
    return chunks

def main():
    # Set up command line arguments
    parser = argparse.ArgumentParser(description="Stream MP3 to Bodet PSA System")
    parser.add_argument("mp3_file", help="Path to MP3 file to stream")
    parser.add_argument("-a", "--addr", default=DEFAULT_MULTICAST_ADDR, 
                        help=f"Multicast address (default: {DEFAULT_MULTICAST_ADDR})")
    parser.add_argument("-p", "--port", type=int, default=DEFAULT_PORT,
                        help=f"UDP port (default: {DEFAULT_PORT})")
    parser.add_argument("-z", "--zone", default=DEFAULT_ZONE_INFO, 
                        help=f"Hex zone info (default: Zone 1)")
    parser.add_argument("-n", "--no-duplicates", action="store_true",
                        help="Don't send duplicate packets (default: send duplicates)")
    parser.add_argument("-c", "--chunk-size", type=int, default=DEFAULT_CHUNK_SIZE,
                        help=f"MP3 chunk size in bytes (default: {DEFAULT_CHUNK_SIZE})")
    parser.add_argument("-q", "--quality", choices=["high", "low"], default="high",
                        help="Audio quality preset: high (128kbps, 48kHz) or low (40kbps, 32kHz) (default: high)")
    parser.add_argument("-m", "--include-metadata", action="store_true",
                        help="Include metadata in MP3 stream (default: no metadata)")
    parser.add_argument("-s", "--sample-format", choices=["s16", "s24", "s32"], default=None,
                        help="Sample format to use for transcoding (default: s16)")
    parser.add_argument("--show-packet-length", action="store_true",
                        help="Show packet length information during streaming")
    parser.add_argument("-t", "--timing-factor", type=float, default=DEFAULT_TIMING_FACTOR,
                        help=f"Timing adjustment factor (lower = faster playback, default: {DEFAULT_TIMING_FACTOR})")
    parser.add_argument("--fixed-stream-id", type=int, 
                        help="Use a fixed stream ID instead of a random one (0-65535)")
    
    args = parser.parse_args()
    
    # Stream the MP3 file
    stream_mp3_to_psa(
        args.mp3_file,
        args.addr,
        args.port,
        args.zone,
        not args.no_duplicates,
        args.chunk_size,
        args.quality,
        args.include_metadata,
        args.sample_format,
        args.show_packet_length,
        args.timing_factor,
        args.fixed_stream_id  # Pass fixed_stream_id parameter
    )

# Remove duplicate main() call
if __name__ == "__main__":
    main()