sonusai 0.15.9__py3-none-any.whl → 0.16.1__py3-none-any.whl

sonusai/__init__.py

@@ -5,6 +5,25 @@ from os.path import dirname
 __version__ = metadata.version(__package__)
 BASEDIR = dirname(__file__)
 
+commands_doc = """
+   audiofe                      Audio front end
+   calc_metric_spenh            Run speech enhancement and analysis
+   doc                          Documentation
+   genft                        Generate feature and truth data
+   genmix                       Generate mixture and truth data
+   genmixdb                     Generate a mixture database
+   gentcst                      Generate target configuration from a subdirectory tree
+   lsdb                         List information about a mixture database
+   mkmanifest                   Make ASR manifest JSON file
+   mkwav                        Make WAV files from a mixture database
+   onnx_predict                 Run ONNX predict on a trained model
+   plot                         Plot mixture data
+   post_spenh_targetf           Run post-processing for speech enhancement targetf data
+   summarize_metric_spenh       Summarize speech enhancement and analysis results
+   tplot                        Plot truth data
+   vars                         List custom SonusAI variables
+"""
+
 # create logger
 logger = logging.getLogger('sonusai')
 logger.setLevel(logging.DEBUG)
@@ -21,7 +40,7 @@ class SonusAIError(Exception):
 
 
 # create file handler
-def create_file_handler(filename: str):
+def create_file_handler(filename: str) -> None:
     fh = logging.FileHandler(filename=filename, mode='w')
     fh.setLevel(logging.DEBUG)
     fh.setFormatter(formatter)
@@ -29,7 +48,7 @@ def create_file_handler(filename: str):
 
 
 # update console handler
-def update_console_handler(verbose: bool):
+def update_console_handler(verbose: bool) -> None:
     if not verbose:
         logger.removeHandler(console_handler)
         console_handler.setLevel(logging.INFO)
@@ -37,14 +56,17 @@ def update_console_handler(verbose: bool):
 
 
 # write initial log message
-def initial_log_messages(name: str):
+def initial_log_messages(name: str, subprocess: str = None) -> None:
     from datetime import datetime
     from getpass import getuser
     from os import getcwd
     from socket import gethostname
     from sys import argv
 
-    logger.info(f'SonusAI {__version__}')
+    if subprocess is None:
+        logger.info(f'SonusAI {__version__}')
+    else:
+        logger.info(f'SonusAI {subprocess}')
     logger.info(f'{name}')
     logger.info('')
     logger.debug(f'Host:      {gethostname()}')
@@ -53,3 +75,13 @@ def initial_log_messages(name: str):
     logger.debug(f'Date:      {datetime.now()}')
     logger.debug(f'Command:   {" ".join(argv)}')
     logger.debug('')
+
+
+def commands_list(doc: str = commands_doc) -> list[str]:
+    lines = doc.split('\n')
+    commands = []
+    for line in lines:
+        command = line.strip().split(' ').pop(0)
+        if command:
+            commands.append(command)
+    return commands

sonusai/audiofe.py

@@ -24,6 +24,10 @@ audiofe_capture_<TIMESTAMP>.wav.
 If a model is specified, run prediction on audio data from this model. Then compute the inverse transform of the
 prediction result and save to audiofe_predict_<TIMESTAMP>.wav.
 
+Also, if a model is specified, save plots of the capture data (time-domain signal and feature) to
+audiofe_capture_<TIMESTAMP>.png and predict data (time-domain signal and feature) to
+audiofe_predict_<TIMESTAMP>.png.
+
 If an ASR is specified, run ASR on the captured audio and print the results. In addition, if a model was also specified,
 run ASR on the predict audio and print the results.
 
@@ -31,41 +35,32 @@ If the debug option is enabled, write capture audio, feature, reconstruct audio,
 audiofe_<TIMESTAMP>.h5.
 
 """
-from os.path import exists
-from select import select
-from sys import stdin
-from typing import Any
+import signal
 
-import h5py
 import numpy as np
-import pyaudio
-import torch
-from docopt import docopt
-from docopt import printable_usage
-
-import sonusai
-from sonusai import create_file_handler
-from sonusai import initial_log_messages
-from sonusai import logger
-from sonusai import update_console_handler
+
 from sonusai.mixture import AudioT
-from sonusai.mixture import CHANNEL_COUNT
-from sonusai.mixture import SAMPLE_RATE
-from sonusai.mixture import get_audio_from_feature
-from sonusai.mixture import get_feature_from_audio
-from sonusai.mixture import read_audio
-from sonusai.utils import calc_asr
-from sonusai.utils import create_timestamp
-from sonusai.utils import get_input_device_index_by_name
-from sonusai.utils import get_input_devices
-from sonusai.utils import import_keras_model
-from sonusai.utils import trim_docstring
-from sonusai.utils import write_wav
+
+
+def signal_handler(_sig, _frame):
+    import sys
+
+    from sonusai import logger
+
+    logger.info('Canceled due to keyboard interrupt')
+    sys.exit(1)
+
+
+signal.signal(signal.SIGINT, signal_handler)
 
 
 def main() -> None:
+    from docopt import docopt
+
+    import sonusai
+    from sonusai.utils import trim_docstring
+
     args = docopt(trim_docstring(__doc__), version=sonusai.__version__, options_first=True)
-    ts = create_timestamp()
 
     verbose = args['--verbose']
     length = float(args['--length'])
@@ -77,8 +72,34 @@ def main() -> None:
     debug = args['--debug']
     show = args['--show']
 
-    capture_name = f'audiofe_capture_{ts}.wav'
-    predict_name = f'audiofe_predict_{ts}.wav'
+    from os.path import exists
+
+    import h5py
+    import pyaudio
+    import torch
+    from docopt import printable_usage
+    from sonusai_torchl.utils import load_torchl_ckpt_model
+
+    from sonusai import create_file_handler
+    from sonusai import initial_log_messages
+    from sonusai import logger
+    from sonusai import update_console_handler
+    from sonusai.mixture import SAMPLE_RATE
+    from sonusai.mixture import get_audio_from_feature
+    from sonusai.mixture import get_feature_from_audio
+    from sonusai.utils import calc_asr
+    from sonusai.utils import create_timestamp
+    from sonusai.utils import get_input_devices
+    from sonusai.utils import trim_docstring
+    from sonusai.utils import write_wav
+
+    ts = create_timestamp()
+    capture_name = f'audiofe_capture_{ts}'
+    capture_wav = capture_name + '.wav'
+    capture_png = capture_name + '.png'
+    predict_name = f'audiofe_predict_{ts}'
+    predict_wav = predict_name + '.wav'
+    predict_png = predict_name + '.png'
     h5_name = f'audiofe_{ts}.h5'
 
     if model_name is not None and ckpt_name is None:
@@ -109,9 +130,9 @@ def main() -> None:
             logger.exception(e)
             return
 
-    write_wav(capture_name, capture_audio, SAMPLE_RATE)
+    write_wav(capture_wav, capture_audio, SAMPLE_RATE)
     logger.info('')
-    logger.info(f'Wrote capture audio with shape {capture_audio.shape} to {capture_name}')
+    logger.info(f'Wrote capture audio with shape {capture_audio.shape} to {capture_wav}')
     if debug:
         with h5py.File(h5_name, 'a') as f:
             if 'capture_audio' in f:
@@ -124,9 +145,13 @@ def main() -> None:
         logger.info(f'Capture audio ASR: {capture_asr}')
 
     if model_name is not None:
-        model = load_model(model_name=model_name, ckpt_name=ckpt_name)
+        model = load_torchl_ckpt_model(model_name=model_name, ckpt_name=ckpt_name)
+        model.eval()
 
         feature = get_feature_from_audio(audio=capture_audio, feature_mode=model.hparams.feature)
+        save_figure(capture_png, capture_audio, feature)
+        logger.info(f'Wrote capture plots to {capture_png}')
+
         if debug:
             with h5py.File(h5_name, 'a') as f:
                 if 'feature' in f:
@@ -134,22 +159,9 @@ def main() -> None:
                 f.create_dataset('feature', data=feature)
             logger.info(f'Wrote feature with shape {feature.shape} to {h5_name}')
 
-        # if debug:
-        #     reconstruct_name = f'audiofe_reconstruct_{ts}.wav'
-        #     reconstruct_audio = get_audio_from_feature(feature=feature, feature_mode=model.hparams.feature)
-        #     samples = min(len(capture_audio), len(reconstruct_audio))
-        #     max_err = np.max(np.abs(capture_audio[:samples] - reconstruct_audio[:samples]))
-        #     logger.info(f'Maximum error between capture and reconstruct: {max_err}')
-        #     write_wav(reconstruct_name, reconstruct_audio, SAMPLE_RATE)
-        #     logger.info(f'Wrote reconstruct audio with shape {reconstruct_audio.shape} to {reconstruct_name}')
-        #     with h5py.File(h5_name, 'a') as f:
-        #         if 'reconstruct_audio' in f:
-        #             del f['reconstruct_audio']
-        #         f.create_dataset('reconstruct_audio', data=reconstruct_audio)
-        #     logger.info(f'Wrote reconstruct audio with shape {reconstruct_audio.shape} to {h5_name}')
-
         with torch.no_grad():
-            predict = model(torch.tensor(feature))
+            # model wants batch x timesteps x feature_parameters
+            predict = model(torch.tensor(feature).permute((1, 0, 2))).permute(1, 0, 2).numpy()
         if debug:
             with h5py.File(h5_name, 'a') as f:
                 if 'predict' in f:
@@ -157,9 +169,9 @@ def main() -> None:
                 f.create_dataset('predict', data=predict)
             logger.info(f'Wrote predict with shape {predict.shape} to {h5_name}')
 
-        predict_audio = get_audio_from_feature(feature=predict.numpy(), feature_mode=model.hparams.feature)
-        write_wav(predict_name, predict_audio, SAMPLE_RATE)
-        logger.info(f'Wrote predict audio with shape {predict_audio.shape} to {predict_name}')
+        predict_audio = get_audio_from_feature(feature=predict, feature_mode=model.hparams.feature)
+        write_wav(predict_wav, predict_audio, SAMPLE_RATE)
+        logger.info(f'Wrote predict audio with shape {predict_audio.shape} to {predict_wav}')
         if debug:
             with h5py.File(h5_name, 'a') as f:
                 if 'predict_audio' in f:
@@ -167,69 +179,26 @@ def main() -> None:
                 f.create_dataset('predict_audio', data=predict_audio)
             logger.info(f'Wrote predict audio with shape {predict_audio.shape} to {h5_name}')
 
+        save_figure(predict_png, predict_audio, predict)
+        logger.info(f'Wrote predict plots to {predict_png}')
+
         if asr_name is not None:
             predict_asr = calc_asr(predict_audio, engine=asr_name, whisper_model_name=whisper_name).text
             logger.info(f'Predict audio ASR: {predict_asr}')
 
 
-def load_model(model_name: str, ckpt_name: str) -> Any:
-    batch_size = 1
-    timesteps = 0
-
-    # Load checkpoint first to get hparams if available
-    try:
-        checkpoint = torch.load(ckpt_name, map_location=lambda storage, loc: storage)
-    except Exception as e:
-        logger.exception(f'Error: could not load checkpoint from {ckpt_name}: {e}')
-        raise SystemExit(1)
-
-    # Import model definition file
-    logger.info(f'Importing {model_name}')
-    litemodule = import_keras_model(model_name)
-
-    if 'hyper_parameters' in checkpoint:
-        logger.info(f'Found checkpoint file with hyper-parameters')
-        hparams = checkpoint['hyper_parameters']
-        if hparams['batch_size'] != batch_size:
-            logger.info(
-                f'Overriding model default batch_size of {hparams["batch_size"]} with batch_size of {batch_size}')
-            hparams["batch_size"] = batch_size
-
-        if hparams['timesteps'] != 0 and timesteps == 0:
-            timesteps = hparams['timesteps']
-            logger.warning(f'Using model default timesteps of {timesteps}')
-
-        logger.info(f'Building model with {len(hparams)} total hparams')
-        try:
-            model = litemodule.MyHyperModel(**hparams)
-        except Exception as e:
-            logger.exception(f'Error: model build (MyHyperModel) in {model_name} failed: {e}')
-            raise SystemExit(1)
-    else:
-        logger.info(f'Found checkpoint file with no hyper-parameters')
-        logger.info(f'Building model with defaults')
-        try:
-            tmp = litemodule.MyHyperModel()
-        except Exception as e:
-            logger.exception(f'Error: model build (MyHyperModel) in {model_name} failed: {e}')
-            raise SystemExit(1)
-
-        if tmp.batch_size != batch_size:
-            logger.info(f'Overriding model default batch_size of {tmp.batch_size} with batch_size of {batch_size}')
-
-        if tmp.timesteps != 0 and timesteps == 0:
-            timesteps = tmp.timesteps
-            logger.warning(f'Using model default timesteps of {timesteps}')
-
-        model = litemodule.MyHyperModel(timesteps=timesteps, batch_size=batch_size)
+def get_frames_from_device(input_name: str | None, length: float, chunk: int = 1024) -> AudioT:
+    from select import select
+    from sys import stdin
 
-    logger.info(f'Loading weights from {ckpt_name}')
-    model.load_state_dict(checkpoint["state_dict"])
-    model.eval()
-    return model
+    import pyaudio
 
+    from sonusai import logger
+    from sonusai.mixture import CHANNEL_COUNT
+    from sonusai.mixture import SAMPLE_RATE
+    from sonusai.utils import get_input_device_index_by_name
+    from sonusai.utils import get_input_devices
 
-def get_frames_from_device(input_name: str | None, length: float, chunk: int = 1024) -> AudioT:
     p = pyaudio.PyAudio()
 
     input_devices = get_input_devices(p)
@@ -280,6 +249,10 @@ def get_frames_from_device(input_name: str | None, length: float, chunk: int = 1
 
 
 def get_frames_from_file(input_name: str, length: float) -> AudioT:
+    from sonusai import logger
+    from sonusai.mixture import SAMPLE_RATE
+    from sonusai.mixture import read_audio
+
     logger.info(f'Capturing from {input_name}')
     frames = read_audio(input_name)
     if length != -1:
@@ -289,5 +262,37 @@ def get_frames_from_file(input_name: str, length: float) -> AudioT:
     return frames
 
 
+def save_figure(name: str, audio: np.ndarray, feature: np.ndarray) -> None:
+    import matplotlib.pyplot as plt
+    from scipy.interpolate import CubicSpline
+
+    from sonusai.mixture import SAMPLE_RATE
+    from sonusai.utils import unstack_complex
+
+    spectrum = 20 * np.log(np.abs(np.squeeze(unstack_complex(feature)).transpose()))
+    frames = spectrum.shape[1]
+    samples = (len(audio) // frames) * frames
+    length_in_s = samples / SAMPLE_RATE
+    interp = samples // frames
+
+    ts = np.arange(0.0, length_in_s, interp / SAMPLE_RATE)
+    t = np.arange(0.0, length_in_s, 1 / SAMPLE_RATE)
+
+    spectrum = CubicSpline(ts, spectrum, axis=-1)(t)
+
+    fig, (ax1, ax2) = plt.subplots(nrows=2)
+    ax1.set_title(name)
+    ax1.plot(t, audio[:samples])
+    ax1.set_ylabel('Signal')
+    ax1.set_xlim(0, length_in_s)
+    ax1.set_ylim(-1, 1)
+
+    ax2.imshow(spectrum, origin='lower', aspect='auto')
+    ax2.set_xticks([])
+    ax2.set_ylabel('Feature')
+
+    plt.savefig(name, dpi=300)
+
+
 if __name__ == '__main__':
     main()

sonusai/calc_metric_spenh.py

@@ -60,6 +60,7 @@ Metric and extraction data are written into prediction location PLOC as separate
 Inputs:
 
 """
+import signal
 from dataclasses import dataclass
 from typing import Optional
 
@@ -67,14 +68,24 @@ import matplotlib
 import matplotlib.pyplot as plt
 import numpy as np
 import pandas as pd
-
-from sonusai import logger
 from sonusai.mixture import AudioF
 from sonusai.mixture import AudioT
 from sonusai.mixture import Feature
 from sonusai.mixture import MixtureDatabase
 from sonusai.mixture import Predict
 
+
+def signal_handler(_sig, _frame):
+    import sys
+
+    from sonusai import logger
+
+    logger.info('Canceled due to keyboard interrupt')
+    sys.exit(1)
+
+
+signal.signal(signal.SIGINT, signal_handler)
+
 matplotlib.use('SVG')
 
 
@@ -758,13 +769,18 @@ def _process_mixture(mixid: int) -> tuple[pd.DataFrame, pd.DataFrame]:
             predict = stack_complex(predict)
 
     # 2) Collect true target, noise, mixture data, trim to predict size if needed
-    target = mixdb.mixture_target(mixid)
-    target_f = mixdb.mixture_target_f(mixid, target=target)
-    noise = mixdb.mixture_noise(mixid)
-    noise_f = mixdb.mixture_noise_f(mixid, noise=noise)
-    mixture = mixdb.mixture_mixture(mixid, target=target, noise=noise)
+    tmp = mixdb.mixture_targets(mixid)  # targets is list of pre-IR and pre-specaugment targets
+    target_f = mixdb.mixture_targets_f(mixid, targets=tmp)[0]
+    target = tmp[0]
+    mixture = mixdb.mixture_mixture(mixid)  # note: gives full reverberated/distorted target, but no specaugment
+    # noise_wodist = mixdb.mixture_noise(mixid)            # noise without specaugment and distortion
+    # noise_wodist_f = mixdb.mixture_noise_f(mixid, noise=noise_wodist)
+    noise = mixture - target  # has time-domain distortion (ir,etc.) but does not have specaugment
+    # noise_f = mixdb.mixture_noise_f(mixid, noise=noise)
+    segsnr_f = mixdb.mixture_segsnr(mixid, target=target, noise=noise)  # note: uses pre-IR, pre-specaug audio
     mixture_f = mixdb.mixture_mixture_f(mixid, mixture=mixture)
-    segsnr_f = mixdb.mixture_segsnr(mixid, target=target, noise=noise)
+    noise_f = mixture_f - target_f  # true noise in freq domain includes specaugment and time-domain ir,distortions
+    # segsnr_f = mixdb.mixture_segsnr(mixid, target=target, noise=noise)
     segsnr_f[segsnr_f == inf] = 7.944e8  # 99db
     segsnr_f[segsnr_f == -inf] = 1.258e-10  # -99db
     # need to use inv-tf to match #samples & latency shift properties of predict inv tf
@@ -920,8 +936,9 @@ def _process_mixture(mixid: int) -> tuple[pd.DataFrame, pd.DataFrame]:
                           'NLERR': lerr_n_frame,
                           'SPD':   phd_frame})
     metr2 = metr2.describe()  # Use pandas stat function
-    metr2['SSNR'][1:] = metr2['SSNR'][1:].apply(
-        lambda x: 10 * np.log10(x + 1.01e-10))  # Change SSNR stats to dB, except count
+    # Change SSNR stats to dB, except count.  SSNR is index 0, pandas requires using iloc
+    # metr2['SSNR'][1:] = metr2['SSNR'][1:].apply(lambda x: 10 * np.log10(x + 1.01e-10))
+    metr2.iloc[1:, 0] = metr2['SSNR'][1:].apply(lambda x: 10 * np.log10(x + 1.01e-10))
     # create a single row in multi-column header
     new_labels = pd.MultiIndex.from_product([metr2.columns,
                                              ['Avg', 'Min', 'Med', 'Max', 'Std']],
@@ -978,11 +995,11 @@ def _process_mixture(mixid: int) -> tuple[pd.DataFrame, pd.DataFrame]:
         plot_fname = base_name + '_metric_spenh.pdf'
 
         # Reshape feature to eliminate overlap redundancy for easier to understand spectrogram view
-        # Original size (frames, stride, feature_parameters), decimates in stride dimension only if step is > 1
-        # Reshape to get frames*decimated_stride, feature_parameters
+        # Original size (frames, stride, num_bands), decimates in stride dimension only if step is > 1
+        # Reshape to get frames*decimated_stride, num_bands
         step = int(mixdb.feature_samples / mixdb.feature_step_samples)
         if feature.ndim != 3:
-            raise SonusAIError(f'feature does not have 3 dimensions: frames, stride, feature_parameters')
+            raise SonusAIError(f'feature does not have 3 dimensions: frames, stride, num_bands')
 
         # for feature cn*00n**
         feat_sgram = unstack_complex(feature)
@@ -1166,7 +1183,7 @@ def main():
     # Individual mixtures use pandas print, set precision to 2 decimal places
     # pd.set_option('float_format', '{:.2f}'.format)
     progress = tqdm(total=len(mixids), desc='calc_metric_spenh')
-    all_metrics_tables = pp_tqdm_imap(_process_mixture, mixids, progress=progress, num_cpus=None)
+    all_metrics_tables = pp_tqdm_imap(_process_mixture, mixids, progress=progress, num_cpus=8)
     progress.close()
 
     all_metrics_table_1 = pd.concat([item[0] for item in all_metrics_tables])
@@ -1192,6 +1209,7 @@ def main():
         if ~np.isnan(tmp.iloc[0].to_numpy()[0]).any():
             mtab_snr_summary_em = pd.concat([mtab_snr_summary_em, tmp])
 
+    mtab_snr_summary = mtab_snr_summary.sort_values(by=['MXSNR'], ascending=False)
     # Correct percentages in snr summary table
     mtab_snr_summary['PESQi%'] = 100 * (mtab_snr_summary['PESQ'] - mtab_snr_summary['MXPESQ']) / np.maximum(
         mtab_snr_summary['MXPESQ'], 0.01)
@@ -1202,9 +1220,11 @@ def main():
             else:
                 mtab_snr_summary['WERi%'].iloc[i] = -999.0
         else:
-            mtab_snr_summary['WERi%'].iloc[i] = 100 * (mtab_snr_summary['MXWER'].iloc[i] -
-                                                       mtab_snr_summary['WER'].iloc[i]) / \
-                                                mtab_snr_summary['MXWER'].iloc[i]
+            if ~np.isnan(mtab_snr_summary['WER'].iloc[i]) and ~np.isnan(mtab_snr_summary['MXWER'].iloc[i]):
+                # update WERi% in 6th col
+                mtab_snr_summary.iloc[i, 6] = 100 * (mtab_snr_summary['MXWER'].iloc[i] -
+                                                     mtab_snr_summary['WER'].iloc[i]) / \
+                                              mtab_snr_summary['MXWER'].iloc[i]
 
     # Calculate avg metrics over all mixtures except -99
     all_mtab1_sorted_nom99 = all_mtab1_sorted[all_mtab1_sorted.MXSNR != -99]
@@ -1317,8 +1337,4 @@ def main():
 
 
 if __name__ == '__main__':
-    try:
-        main()
-    except KeyboardInterrupt:
-        logger.info('Canceled due to keyboard interrupt')
-        exit()
+    main()

sonusai/genft.py

@@ -23,14 +23,26 @@ Outputs the following to the mixture database directory:
     genft.log
 
 """
+import signal
 from dataclasses import dataclass
 
-from sonusai import logger
 from sonusai.mixture import GenFTData
 from sonusai.mixture import GeneralizedIDs
 from sonusai.mixture import MixtureDatabase
 
 
+def signal_handler(_sig, _frame):
+    import sys
+
+    from sonusai import logger
+
+    logger.info('Canceled due to keyboard interrupt')
+    sys.exit(1)
+
+
+signal.signal(signal.SIGINT, signal_handler)
+
+
 @dataclass
 class MPGlobal:
     mixdb: MixtureDatabase = None
@@ -123,6 +135,7 @@ def main() -> None:
 
     from sonusai import create_file_handler
     from sonusai import initial_log_messages
+    from sonusai import logger
     from sonusai import update_console_handler
     from sonusai.mixture import check_audio_files_exist
     from sonusai.utils import human_readable_size
@@ -177,8 +190,4 @@ def main() -> None:
 
 
 if __name__ == '__main__':
-    try:
-        main()
-    except KeyboardInterrupt:
-        logger.info('Canceled due to keyboard interrupt')
-        raise SystemExit(0)
+    main()

sonusai/genmix.py

@@ -27,14 +27,26 @@ Outputs the following to the mixture database directory:
     <id>.txt
     genmix.log
 """
+import signal
 from dataclasses import dataclass
 
-from sonusai import logger
 from sonusai.mixture import GenMixData
 from sonusai.mixture import GeneralizedIDs
 from sonusai.mixture import MixtureDatabase
 
 
+def signal_handler(_sig, _frame):
+    import sys
+
+    from sonusai import logger
+
+    logger.info('Canceled due to keyboard interrupt')
+    sys.exit(1)
+
+
+signal.signal(signal.SIGINT, signal_handler)
+
+
 @dataclass
 class MPGlobal:
     mixdb: MixtureDatabase = None
@@ -210,8 +222,4 @@ def main() -> None:
 
 
 if __name__ == '__main__':
-    try:
-        main()
-    except KeyboardInterrupt:
-        logger.info('Canceled due to keyboard interrupt')
-        raise SystemExit(0)
+    main()

sonusai/genmixdb.py

@@ -112,13 +112,25 @@ targets:
 will find all .wav files in the specified directories and process them as targets.
 
 """
+import signal
 from dataclasses import dataclass
 
-from sonusai import logger
 from sonusai.mixture import Mixture
 from sonusai.mixture import MixtureDatabase
 
 
+def signal_handler(_sig, _frame):
+    import sys
+
+    from sonusai import logger
+
+    logger.info('Canceled due to keyboard interrupt')
+    sys.exit(1)
+
+
+signal.signal(signal.SIGINT, signal_handler)
+
+
 @dataclass
 class MPGlobal:
     mixdb: MixtureDatabase = None
@@ -225,7 +237,7 @@ def genmixdb(location: str,
     if logging:
         logger.info('Collecting impulse responses')
 
-    impulse_response_files = get_impulse_response_files(config, show_progress=show_progress)
+    impulse_response_files = get_impulse_response_files(config)
 
     populate_impulse_response_file_table(location, impulse_response_files, test)
 
@@ -509,8 +521,4 @@ def main() -> None:
 
 
 if __name__ == '__main__':
-    try:
-        main()
-    except KeyboardInterrupt:
-        logger.info('Canceled due to keyboard interrupt')
-        raise SystemExit(0)
+    main()

sonusai/gentcst.py

@@ -44,10 +44,21 @@ Outputs:
     gentcst.log
 
 """
+import signal
 from dataclasses import dataclass
 from typing import Optional
 
-from sonusai import logger
+
+def signal_handler(_sig, _frame):
+    import sys
+
+    from sonusai import logger
+
+    logger.info('Canceled due to keyboard interrupt')
+    sys.exit(1)
+
+
+signal.signal(signal.SIGINT, signal_handler)
 
 CONFIG_FILE = 'config.yml'
 
@@ -621,8 +632,4 @@ def main() -> None:
 
 
 if __name__ == '__main__':
-    try:
-        main()
-    except KeyboardInterrupt:
-        logger.info('Canceled due to keyboard interrupt')
-        raise SystemExit(0)
+    main()