tictacsync 0.1a10__tar.gz → 0.1a12__tar.gz

tictacsync-0.1a12/PKG-INFO

@@ -0,0 +1,87 @@
+Metadata-Version: 2.1
+Name: tictacsync
+Version: 0.1a12
+Summary: command for syncing audio video recordings
+Home-page: https://sr.ht/~proflutz/TicTacSync/
+Author: Raymond Lutz
+Author-email: lutzrayblog@mac.com
+Classifier: Development Status :: 2 - Pre-Alpha
+Classifier: Environment :: Console
+Classifier: Intended Audience :: End Users/Desktop
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: MacOS
+Classifier: Operating System :: Microsoft :: Windows
+Classifier: Operating System :: POSIX
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Multimedia :: Sound/Audio
+Classifier: Topic :: Utilities
+Classifier: Topic :: Multimedia :: Sound/Audio :: Capture/Recording
+Classifier: Topic :: Multimedia :: Video :: Non-Linear Editor
+Requires-Python: >=3.10
+Description-Content-Type: text/markdown
+License-File: LICENSE
+
+# tictacsync
+
+## Warning: this is at pre-alpha stage
+
+Unfinished sloppy code ahead, but should run without errors. Some functionalities are still missing. Don't run the code without parental supervision. Suggestions and enquiries are welcome via the [lists hosted on sourcehut](https://sr.ht/~proflutz/TicTacSync/lists).
+
+## Description
+
+`tictacsync` is a python script to sync audio and video files shot
+with [dual system sound](https://www.learnlightandsound.com/blog/2017/2/23/how-to-record-sound-for-video-dual-systemsync-sound)  using a specific hardware timecode generator
+called [Tic Tac Sync](https://tictacsync.org). The timecode is named YaLTC for *yet
+another longitudinal time code* and should be recorded on a scratch
+track on each device for the syncing to be performed, later in _postprod_ before editing.
+
+## Status
+
+`tictacsync`  scans for audio video files and displays their starting time and then merges overlapping audio and video recordings. Multicam syncing with one stereo audio recorder has been tested (spring 2023, [see demo](https://youtu.be/pklTSTi7cqs)). Multi audio recorders coming soon... 
+
+
+## Installation
+
+This uses the [python interpreter](https://www.python.org/downloads/) and multiple packages (so you need python 3 + pip). Also, you need to install two non-python command line executables: [ffmpeg](https://windowsloop.com/install-ffmpeg-windows-10/) and [sox](https://sourceforge.net/projects/sox/files/). Make sure those are _accessible through your `PATH` system environment variable_.
+Then pip install the syncing program:
+
+
+    pip install tictacsync
+
+
+This should install python dependencies _and_ the `tictacsync` command.
+## Usage
+
+Download some sample files [here](https://tictacsync.org/sampleFiles.zip), unzip and run 
+
+    tictacsync sampleFiles
+The resulting synced videos will be in a subfolder named `tictacsynced`. For a very verbose output add the `-v` flag:
+
+    tictacsync -v sampleFiles
+
+When the argument is an unique media file (not a directory), no syncing will occur but the decoded starting time will be printed to stdout:
+	
+	tictacsync sampleFiles/canon24fps01.MOV
+
+	Recording started at 2023-04-23 01:09:08.1605 UTC
+	true sample rate: 48000.545 Hz
+	first sync at 37414 samples
+	N.B.: all results are precise to the displayed digits!
+
+
+When called with the `-p` flag, zoomable plots will be produced for diagnostic purpose (close the plotting window for the 2nd one) and the decoded starting time will be output to stdin:
+
+    tictacsync -p sampleFiles/canon24fps01.MOV
+
+Typical first plot produced :
+
+![word](https://mamot.fr/system/media_attachments/files/110/279/794/002/305/269/original/0198908c6eb5c592.png)
+
+Typical second plot produced (note the 34 [FSK](https://en.wikipedia.org/wiki/Frequency-shift_keying) encoded bits `0010111101001111100110000110010000`):
+![slicing](https://mamot.fr/system/media_attachments/files/110/279/794/021/372/766/original/6ec62bb417115f52.png)
+
+
+<!-- To run some tests, from top level `git cloned` dir:
+
+    cd tictacsync ; python -m pytest
+ Yes, the coverage is low. -->

{tictacsync-0.1a10 → tictacsync-0.1a12}/setup.py

@@ -20,14 +20,18 @@ setup(
         'matplotlib>=3.7.1',
         'numpy>=1.24.3',
         'rich>=10.12.0',
-        'lmfit >= 1.0.2',
+        'lmfit',
         'scipy>=1.10.1',
         'scikit-learn==1.2.2'], 
-    python_requires='>=3.7',
+    python_requires='>=3.10',
     entry_points = {
-        "console_scripts": ['tictacsync = tictacsync.entry:main']
+        "console_scripts": [
+        'tictacsync = tictacsync.entry:main',
+        'LTCcheck = tictacsync.LTCcheck:main',
+        'multi2polywav = tictacsync.multi2polywav:main',
+        ]
         },
-    version = '0.1a10',
+    version = '0.1a12',
     description = "command for syncing audio video recordings",
     long_description_content_type='text/markdown',
     long_description = long_descr,

tictacsync-0.1a12/tictacsync/LTCcheck.py

@@ -0,0 +1,394 @@
+print('Loading modules')
+import subprocess, io
+import argparse, os, sys, ffmpeg
+from loguru import logger
+from pathlib import Path
+from scipy.io import wavfile
+import numpy as np
+import matplotlib.pyplot as plt
+from rich.progress import track, Progress
+from pprint import pprint
+from collections import deque
+import wave
+try:
+    from . import yaltc
+except:
+    import yaltc
+try:
+    from . import device_scanner
+except:
+    import device_scanner
+
+# LEVELMODE = 'over_noise_silence'
+LEVELMODE = 'mean_silence_AFSK'
+
+OFFSET_NXT_PULSE = 50 # samples
+LENGTH_EXTRACT = int(14e-3 * 96000) # samples max freq
+
+logger.level("DEBUG", color="<yellow>")
+logger.remove()
+# logger.add(sys.stdout, filter="tictacsync.LTCcheck")
+# logger.add(sys.stdout, filter="tictacsync.yaltc")
+
+def ppm(a,b):
+    return 1e6*(max(a,b)/min(a,b)-1)
+
+class TCframe:
+    def __init__(self, string, max_FF):
+        # string is 'HH:MM:SS:FF' or ;|,|.
+        # max_FF is int for max frame number (hence fps-1)
+        string = string.replace('.',':').replace(';',':').replace(',',':')
+        ints = [int(e) for e in string.split(':')]
+        self.HH = ints[0]
+        self.MM = ints[1]
+        self.SS = ints[2]
+        self.FF = ints[3]
+        self.MAXFF = max_FF
+
+    def __repr__(self):
+        # return '%s-%s-%s-%s/%i'%(*self.ints(), self.MAXFF)
+        return '%02i-%02i-%02i-%02i'%self.ints()
+
+    def ints(self):
+        return (self.HH,self.MM,self.SS,self.FF)
+
+    def __eq__(self, other):
+        a,b,c,d = self.ints()
+        h,m,s,f = other.ints()
+        return a==h and b==m and c==s and d==f
+
+    def __sub__(self, tcf2):
+        # H1, M1, S1, F1 = self.ints()
+        # H2, M2, S2, F2 = tcf2.ints()
+        f1 = np.array(self.ints())
+        f2 = np.array(tcf2.ints())
+        HR, MR, SR, FR = f1 - f2
+        if FR < 0:
+            FR += self.MAXFF + 1
+            SR -= 1 # borrow
+        if SR < 0:
+            SR += 60
+            MR -= 1 # borrow
+        if MR < 0:
+            MR += 60
+            HR -= 1 # borrow
+        if HR < 0:
+            HR += 24 # underflow?
+        # logger.debug('%s %s'%(self.ints(), tcf2.ints()))
+        return TCframe('%02i:%02i:%02i:%02i'%(HR,MR,SR,FR), self.MAXFF)
+
+def read_whole_audio_data(path):
+    dryrun = (ffmpeg
+        .input(str(path))
+        .output('pipe:', format='s16le', acodec='pcm_s16le')
+        .get_args())
+    dryrun = ' '.join(dryrun)
+    logger.debug('using ffmpeg-python built args to pipe wav file into numpy array:\nffmpeg %s'%dryrun)
+    try:
+        out, _ = (ffmpeg
+            .input(str(path))
+            .output('pipe:', format='s16le', acodec='pcm_s16le')
+            .global_args("-loglevel", "quiet")
+            .global_args("-nostats")
+            .global_args("-hide_banner")
+            .run(capture_stdout=True))
+        data = np.frombuffer(out, np.int16)
+    except ffmpeg.Error as e:
+        print('error',e.stderr)
+    with wave.open(path, 'rb') as f:
+        samplerate = f.getframerate()
+        n_chan = f.getnchannels()
+    all_channels_data = data.reshape(int(len(data)/n_chan),n_chan).T
+    return all_channels_data   
+
+def find_nearest_fps(value):
+    array = np.asarray([24, 25, 30])
+    idx = (np.abs(array - value)).argmin()
+    return array[idx]
+
+def fps_rel_to_audio(frame_pos, samplerate):
+    _, first_frame_pos = frame_pos[0]
+    _, scnd_last_frame_pos = frame_pos[-2]
+    frame_duration = (scnd_last_frame_pos - first_frame_pos)/len(frame_pos[:-2]) # in audio samples
+    fps = float(samplerate) / frame_duration
+    return fps
+
+# def HHMMSSFF_from_line(line):
+#   line = line.replace('.',':')
+#   line = line.replace(';',':')
+#   ll = line.split()[1].split(':')
+#   return [int(e) for e in ll]
+
+def check_continuity_and_DF(LTC_frames_and_pos):
+    errors = []
+    DF_flag = False
+    oneframe = TCframe('00:00:00:01',None)
+    threeframes = TCframe('00:00:00:03',None)
+    last_two_TC = deque([], maxlen=2)
+    last_two_TC.append(LTC_frames_and_pos[0][0])
+    last_two_TC.append(LTC_frames_and_pos[1][0])
+    for frame, pos in track(LTC_frames_and_pos[2:],
+            description="Checking each frame increment"):
+        last_two_TC.append(frame)
+        past, now = last_two_TC
+        diff = now - past
+        if diff not in [oneframe, threeframes]:
+            errors.append((frame, pos))
+            continue
+        if diff == oneframe:
+            continue
+        if diff == threeframes: 
+            # DF? check if it is 59:xx and minutes are not mult. of tens
+            if past.SS != 59 or now.MM%10 == 0: 
+                errors.append((frame, pos))
+            DF_flag = True
+    return errors, DF_flag
+
+def ltcdump_and_check(file, channel):
+    # returns list of anormal frames, a bool if TC is DF, fps and
+    # a list of tuples (frame => str, sample position in file => int) as
+    # determined by external util ltcdump https://github.com/x42/ltc-tools
+    process_list = ["ltcdump","-c %i"%channel, file]
+    logger.debug('process %s'%process_list)
+    proc = subprocess.Popen(process_list, stdout=subprocess.PIPE)
+    LTC_frames_and_pos = []
+    iter_io = io.TextIOWrapper(proc.stdout, encoding="utf-8")
+    next(iter_io) # ltcdump 1st line: User bits  Timecode   |    Pos. (samples)
+    print()
+    try:
+        next(iter_io) # ltcdump 2nd line: #DISCONTINUITY
+    except StopIteration:
+        print('ltcdump has no output, is channel #%i really LTC?'%channel)
+        quit()
+    old = 0
+    for line in track(iter_io,
+        description='       Parsing ltcdump output'):  # next ones
+        # print(line)
+        if line == '#DISCONTINUITY\n':
+            # print('#DISCONTINUITY!')
+            continue
+        user_bits, HHMMSSFF_str, _, start_sample, end_sample =\
+                line.split()
+        audio_position = int(end_sample)
+        # print(audio_position - old, end=' ')
+        # old = audio_position
+        # audio_position = int(start_sample)
+        tc = HHMMSSFF_str
+        LTC_frames_and_pos.append((tc, audio_position))
+    with wave.open(file, 'rb') as f:
+        samplerate = f.getframerate()
+    fps = fps_rel_to_audio(LTC_frames_and_pos, samplerate)
+    rounded_fps = round(fps)
+    LTC_frames_and_pos = [(TCframe(tc, rounded_fps-1), pos) for tc, pos in LTC_frames_and_pos]
+    errors, DF_flag = check_continuity_and_DF(LTC_frames_and_pos)
+    return errors, DF_flag, fps, LTC_frames_and_pos
+
+def find_pulses(TTC_data, recording):
+    samplerate = recording.true_samplerate
+    i_samplerate = round(samplerate)
+    pulse_position = recording.sync_position
+    logger.debug('first detected pulse %i'%pulse_position)
+    # first_pulse_nbr_of_seconds = int(pulse_position/samplerate)
+    # if first_pulse_nbr_of_seconds > 1:
+    #     pulse_position = pulse_position%i_samplerate # very first pulse in file
+    # print('0 %i'%pulse_position)
+    pulse_position = pulse_position%i_samplerate
+    logger.debug('starting at %i'%pulse_position)
+    second = 0
+    duration = int(recording.get_duration())
+    decoder = recording.decoder
+    pulse_detection_level = decoder._get_pulse_detection_level()
+    logger.debug(' detection level %f'%pulse_detection_level)
+    pulses = []
+    approx_next_pulse = pulse_position
+    skipped_printed = False
+    while second < duration - 1:
+        second += 1
+        approx_next_pulse -= OFFSET_NXT_PULSE
+        start_of_extract = approx_next_pulse
+        sound_extract = TTC_data[start_of_extract:start_of_extract + LENGTH_EXTRACT]
+        abs_signal = abs(sound_extract)
+        detected_point = \
+                np.argmax(abs_signal > pulse_detection_level)
+        old_pulse_position = pulse_position
+        pulse_position = detected_point + start_of_extract
+        diff = pulse_position - old_pulse_position
+        logger.debug('pulse_position %f old_pulse_position %f diff %f'%(pulse_position,
+                        old_pulse_position, diff))
+        if not np.isclose(diff, samplerate, rtol=1e-4):
+            if not skipped_printed:
+                print('\nSkipped: ', end='')
+                skipped_printed = True
+            print('%i, '%(pulse_position), end='')
+            # if diff < samplerate:
+            # else:
+                # print('skipped: samples %i and %i are too far'%(pulse_position, old_pulse_position))
+        else:          
+            pulses.append((second, pulse_position))
+        approx_next_pulse = pulse_position + i_samplerate
+    if skipped_printed:
+        print('\n')
+    return pulses
+
+def main():
+    print('in main()')
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+                "LTC_chan",
+                type=int,
+                # nargs=2,
+                help="LTC channel number"
+                )
+    parser.add_argument(
+                "file_argument",
+                type=str,
+                nargs=1,
+                help="media file"
+                )
+    args = parser.parse_args()
+    # print(args.channels)
+    LTC_chan = args.LTC_chan
+    file_argument = args.file_argument[0]
+    logger.info('args.file_argument: %s'%file_argument)
+    if os.path.isdir(file_argument):
+        print('argument shoud be a media file, not a directory. Bye...')
+        quit()
+    # print(file_argument)
+    if not os.path.exists(file_argument):
+        print('%s does not exist, bye'%file_argument)
+        quit()
+    errors, DF_flag, fps_rel_to_audio, LTC_frames_and_pos = ltcdump_and_check(file_argument, LTC_chan)
+    if errors:
+        print('errors! %s'%errors)
+        print('Some errors in those %i but detected FPS rel to audio is %0.3f%s'%(len(LTC_frames_and_pos),
+                                fps_rel_to_audio, 'DF' if DF_flag else 'NDF'))
+    else:
+        print('\nAll %i frames are sequential and detected FPS rel to audio is %0.3f%s\n'%(len(LTC_frames_and_pos),
+                                fps_rel_to_audio, 'DF' if DF_flag else 'NDF'))
+    # print('trying to decode TTC...')
+    with Progress(transient=True) as progress:
+        task = progress.add_task("trying to decode TTC...")
+        progress.start()
+        m = device_scanner.media_dict_from_path(Path(file_argument))
+        logger.debug('media_dict_from_path %s'%m)
+        recording = yaltc.Recording(m, )
+        logger.debug('Rec %s'%recording)
+        time = recording.get_start_time(progress=progress, task=task)
+    if time == None:
+        print('Start time couldnt be determined')
+    else:
+        audio_samplerate_gps_corrected = recording.true_samplerate
+        audio_error = audio_samplerate_gps_corrected/recording.get_samplerate()
+        gps_corrected_framerate = fps_rel_to_audio*audio_error  
+        print('gps_corrected_framerate',gps_corrected_framerate,audio_error)
+        frac_time = int(time.microsecond / 1e2)
+        d = '%s.%s'%(time.strftime("%Y-%m-%d %H:%M:%S"),frac_time)
+        base = os.path.basename(file_argument)
+        print('%s UTC:%s pulse: %i on chan %i'%(base, d,
+                    recording.sync_position,
+                    recording.YaLTC_channel))
+        print('audio samplerate (gps)', audio_samplerate_gps_corrected)
+        all_channels_data = read_whole_audio_data(file_argument)
+        TTC_data = all_channels_data[recording.YaLTC_channel]
+        sec_and_pulses = find_pulses(TTC_data, recording)
+        secs, pulses = list(zip(*sec_and_pulses))
+        pulses = list(pulses)
+        logger.debug('pulses %s'%pulses)
+        samples_between_UTC_pulses = []
+        for n1, n2 in zip(pulses[1:], pulses):
+            delta = n1 - n2
+            if np.isclose(delta, audio_samplerate_gps_corrected, rtol=1e-3):
+                samples_between_UTC_pulses.append(delta - audio_samplerate_gps_corrected)
+        samples_between_UTC_pulses = np.array(samples_between_UTC_pulses)
+        pulse_length_std = samples_between_UTC_pulses.std()
+        max_min_over_2 = abs(samples_between_UTC_pulses.max() - samples_between_UTC_pulses.min())/2
+        # print(samples_between_UTC_pulses)
+        # print('time is measured with a precision of %f audio samples'%(pulse_length_std))
+        precision = 1e6*max_min_over_2/audio_samplerate_gps_corrected
+        print('Time is measured with a precision of %0.1f audio samples (%0.1f μs)'%(max_min_over_2, precision))
+        frame_duration = 1/fps_rel_to_audio
+        rel_min_error = 100*1e-6*precision/frame_duration
+        print('so LTC syncword jitter less than %0.1f %% wont be detected'%(rel_min_error))
+        # fig, ax = plt.subplots()
+        # n, bins, patches = ax.hist(samples_between_UTC_pulses)
+        # plt.show()
+        # x = range(len(pulses))
+        a, b = np.polyfit(pulses, secs, 1)
+        logger.debug('slope, b = %f %f'%(a,b))
+        # sr_slope = 1/a
+        # print(sr_slope/recording.true_samplerate)
+        coherent_sr = np.isclose(a*audio_samplerate_gps_corrected, 1, rtol=1e-7)
+        logger.debug('samplerates (slope VS rec) are close: %s ratio %f'%(coherent_sr,
+                                    a*audio_samplerate_gps_corrected))
+        if not coherent_sr:
+            print('warning, wav samplerate are incoherent (Rec + Decode VS slope)')
+        def make_sample2time(a, b):
+            return lambda n : a*n + b 
+        sample2time = make_sample2time(a, b)
+        logger.debug('sample2time fct: %s'%sample2time)
+        LTC_samples = [N for _, N in LTC_frames_and_pos]
+        LTC_times = [sample2time(N) for N in LTC_samples]
+        slope_fps, _ = np.polyfit(LTC_times, range(len(LTC_times)), 1)
+        print('slope_fps l329', ppm(slope_fps,24))
+        print('diff slope, ppm',ppm(gps_corrected_framerate, slope_fps))
+        LTC_frame_durations_samples = [a - b for a, b in zip(LTC_samples[1:], LTC_samples)]
+        # print(LTC_frame_durations_samples)
+        frame_duration = 1/fps_rel_to_audio 
+        errors_useconds = [1e6*(frame_duration -(a - b)) for a, b in zip(LTC_times[1:], LTC_times)]
+        # print(errors_useconds)
+        errors_useconds = np.array(errors_useconds)
+        LTC_std = abs(errors_useconds).std()
+        LTC_max_min = abs(errors_useconds.max() - errors_useconds.min())/2
+        # print('Mean frame duration is %i audio samples'%)
+        print('\nhere LTC frame duration varies by %f μs ('%LTC_max_min, end='')
+        print('%0.3fFPS nominal frame duration is %0.0f μs)\n'%(fps_rel_to_audio, 1e6/fps_rel_to_audio))
+        # print(errors_useconds[:200])
+        # audio_sampling_period = 1/samplerate
+        # print(audio_sampling_period)
+        # errors_in_audiosamples = [int(e/audio_sampling_period) for e in errors_seconds]
+        # print(delta_milliseconds)
+        # plt.plot(LTC_times, marker='.', markersize='1',
+        #     linestyle='None', color='black')
+        # plt.show()
+        # print(LTC_times)
+        # fig, ax = plt.subplots()
+
+        # the histogram of the data
+        # print(errors_in_audiosamples)
+        fig, ax = plt.subplots()
+        n, bins, patches = ax.hist(errors_useconds, bins=40)
+        plt.show()
+    quit()
+
+if __name__ == '__main__':
+    main()
+
+#     import matplotlib.pyplot as plt
+# import numpy as np
+
+# rng = np.random.default_rng(19680801)
+
+# # example data
+# mu = 106  # mean of distribution
+# sigma = 17  # standard deviation of distribution
+# x = rng.normal(loc=mu, scale=sigma, size=420)
+
+# num_bins = 42
+
+# fig, ax = plt.subplots()
+
+# # the histogram of the data
+# n, bins, patches = ax.hist(x, num_bins, density=True)
+
+# # add a 'best fit' line
+# y = ((1 / (np.sqrt(2 * np.pi) * sigma)) *
+#      np.exp(-0.5 * (1 / sigma * (bins - mu))**2))
+# ax.plot(bins, y, '--')
+# ax.set_xlabel('Value')
+# ax.set_ylabel('Probability density')
+# ax.set_title('Histogram of normal distribution sample: '
+#              fr'$\mu={mu:.0f}$, $\sigma={sigma:.0f}$')
+
+# # Tweak spacing to prevent clipping of ylabel
+# fig.tight_layout()
+# plt.show()