sonusai 0.19.6__py3-none-any.whl → 0.19.9__py3-none-any.whl

sonusai/mkwav.py

@@ -1,12 +1,13 @@
 """sonusai mkwav
 
-usage: mkwav [-hvtn] [-i MIXID] LOC
+usage: mkwav [-hvtsn] [-i MIXID] LOC
 
 options:
     -h, --help
     -v, --verbose                   Be verbose.
     -i MIXID, --mixid MIXID         Mixture ID(s) to generate. [default: *].
     -t, --target                    Write target file.
+    -s, --targets                   Write targets files.
     -n, --noise                     Write noise file.
 
 The mkwav command creates WAV files from a SonusAI database.
@@ -17,9 +18,10 @@ Inputs:
 
 Outputs the following to the mixture database directory:
     <id>
-        mixture.wav:   mixture
-        target.wav:    target (optional)
-        noise.wav:     noise (optional)
+        mixture.wav:        mixture
+        target.wav:         target (optional)
+        targets<n>.wav:     targets <n> (optional)
+        noise.wav:          noise (optional)
         metadata.txt
     mkwav.log
 
@@ -40,7 +42,7 @@ def signal_handler(_sig, _frame):
 signal.signal(signal.SIGINT, signal_handler)
 
 
-def _process_mixture(m_id: int, location: str, write_target: bool, write_noise: bool) -> None:
+def _process_mixture(m_id: int, location: str, write_target: bool, write_targets: bool, write_noise: bool) -> None:
     from os.path import join
 
     from sonusai.mixture import MixtureDatabase
@@ -50,16 +52,18 @@ def _process_mixture(m_id: int, location: str, write_target: bool, write_noise:
 
     mixdb = MixtureDatabase(location)
 
-    mixture = mixdb.mixture(m_id)
-    location = join(mixdb.location, mixture.name)
+    location = join(mixdb.location, "mixture", mixdb.mixture(m_id).name)
 
     write_audio(name=join(location, "mixture.wav"), audio=float_to_int16(mixdb.mixture_mixture(m_id)))
     if write_target:
         write_audio(name=join(location, "target.wav"), audio=float_to_int16(mixdb.mixture_target(m_id)))
+    if write_targets:
+        for idx, target in enumerate(mixdb.mixture_targets(m_id)):
+            write_audio(name=join(location, f"targets{idx}.wav"), audio=float_to_int16(target))
     if write_noise:
         write_audio(name=join(location, "noise.wav"), audio=float_to_int16(mixdb.mixture_noise(m_id)))
 
-    write_mixture_metadata(mixdb, mixture)
+    write_mixture_metadata(mixdb, m_id)
 
 
 def main() -> None:
@@ -73,6 +77,7 @@ def main() -> None:
     verbose = args["--verbose"]
     mixid = args["--mixid"]
     write_target = args["--target"]
+    write_targets = args["--targets"]
     write_noise = args["--noise"]
     location = args["LOC"]
 
@@ -80,14 +85,12 @@ def main() -> None:
     from functools import partial
     from os.path import join
 
-    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 MixtureDatabase
     from sonusai.mixture import check_audio_files_exist
-    from sonusai.utils import human_readable_size
     from sonusai.utils import par_track
     from sonusai.utils import seconds_to_hms
     from sonusai.utils import track
@@ -103,7 +106,6 @@ def main() -> None:
     mixid = mixdb.mixids_to_list(mixid)
 
     total_samples = mixdb.total_samples(mixid)
-    duration = total_samples / sonusai.mixture.SAMPLE_RATE
 
     logger.info("")
     logger.info(f"Found {len(mixid):,} mixtures to process")
@@ -113,7 +115,13 @@ def main() -> None:
 
     progress = track(total=len(mixid))
     par_track(
-        partial(_process_mixture, location=location, write_target=write_target, write_noise=write_noise),
+        partial(
+            _process_mixture,
+            location=location,
+            write_target=write_target,
+            write_targets=write_targets,
+            write_noise=write_noise,
+        ),
         mixid,
         progress=progress,
     )
@@ -121,13 +129,6 @@ def main() -> None:
 
     logger.info(f"Wrote {len(mixid)} mixtures to {location}")
     logger.info("")
-    logger.info(f"Duration: {seconds_to_hms(seconds=duration)}")
-    logger.info(f"mixture:  {human_readable_size(total_samples * 2, 1)}")
-    if write_target:
-        logger.info(f"target:   {human_readable_size(total_samples * 2, 1)}")
-    if write_noise:
-        logger.info(f"noise:    {human_readable_size(total_samples * 2, 1)}")
-
     end_time = time.monotonic()
     logger.info(f"Completed in {seconds_to_hms(seconds=end_time - start_time)}")
     logger.info("")

sonusai/queries/queries.py

@@ -5,6 +5,10 @@ from sonusai.mixture.datatypes import GeneralizedIDs
 from sonusai.mixture.mixdb import MixtureDatabase
 
 
+def _true_predicate(_: Any) -> bool:
+    return True
+
+
 def get_mixids_from_mixture_field_predicate(
     mixdb: MixtureDatabase,
     field: str,
@@ -20,9 +24,7 @@ def get_mixids_from_mixture_field_predicate(
     mixid_out = mixdb.mixids_to_list(mixids)
 
     if predicate is None:
-
-        def predicate(_: Any) -> bool:
-            return True
+        predicate = _true_predicate
 
     criteria_set = set()
     for m_id in mixid_out:
@@ -70,9 +72,7 @@ def get_mixids_from_truth_configs_field_predicate(
     values = get_all_truth_configs_values_from_field(mixdb, field)
 
     if predicate is None:
-
-        def predicate(_: Any) -> bool:
-            return True
+        predicate = _true_predicate
 
     # Get only values of interest
     values = [value for value in values if predicate(value)]
@@ -118,7 +118,7 @@ def get_all_truth_configs_values_from_field(mixdb: MixtureDatabase, field: str)
                 value = getattr(truth_config, field)
             else:
                 value = getattr(truth_config.config, field, None)
-            if isinstance(value, str):
+            if not isinstance(value, list):
                 value = [value]
             result.extend(value)
 
@@ -164,17 +164,13 @@ def get_mixids_from_snr(
         - keys are the SNRs
         - values are lists of the mixids that match the SNR
     """
-    from typing import Any
-
     mixid_out = mixdb.mixids_to_list(mixids)
 
     # Get all the SNRs
     snrs = [float(snr) for snr in mixdb.all_snrs if not snr.is_random]
 
     if predicate is None:
-
-        def predicate(_: Any) -> bool:
-            return True
+        predicate = _true_predicate
 
     # Get only the SNRs of interest (filter on predicate)
     snrs = [snr for snr in snrs if predicate(snr)]
@@ -201,9 +197,7 @@ def get_mixids_from_class_indices(
     mixid_out = mixdb.mixids_to_list(mixids)
 
     if predicate is None:
-
-        def predicate(_: Any) -> bool:
-            return True
+        predicate = _true_predicate
 
     criteria_set = set()
     for m_id in mixid_out:

sonusai/speech/l2arctic.py

@@ -54,6 +54,7 @@ def load_phonemes(audio: str | os.PathLike[str]) -> list[TimeAlignedType] | None
 
 def _load_ta(audio: str | os.PathLike[str], tier: str) -> list[TimeAlignedType] | None:
     from praatio import textgrid
+    from praatio.utilities.constants import Interval
 
     file = Path(audio).parent.parent / "textgrid" / (Path(audio).stem + ".TextGrid")
     if not os.path.exists(file):
@@ -65,7 +66,8 @@ def _load_ta(audio: str | os.PathLike[str], tier: str) -> list[TimeAlignedType]
 
     entries: list[TimeAlignedType] = []
     for entry in tg.getTier(tier).entries:
-        entries.append(TimeAlignedType(text=entry.label, start=entry.start, end=entry.end))
+        if isinstance(entry, Interval):
+            entries.append(TimeAlignedType(text=entry.label, start=entry.start, end=entry.end))
 
     return entries
 
@@ -79,6 +81,7 @@ def load_annotations(
     :return: A dictionary of a list of TimeAlignedType objects.
     """
     from praatio import textgrid
+    from praatio.utilities.constants import Interval
 
     file = Path(audio).parent.parent / "annotation" / (Path(audio).stem + ".TextGrid")
     if not os.path.exists(file):
@@ -89,7 +92,8 @@ def load_annotations(
     for tier in tg.tierNames:
         entries: list[TimeAlignedType] = []
         for entry in tg.getTier(tier).entries:
-            entries.append(TimeAlignedType(text=entry.label, start=entry.start, end=entry.end))
+            if isinstance(entry, Interval):
+                entries.append(TimeAlignedType(text=entry.label, start=entry.start, end=entry.end))
         result[tier] = entries
 
     return result

sonusai/summarize_metric_spenh.py

@@ -48,7 +48,7 @@ def summarize_metric_spenh(location: str, by: str = "MIXID", reverse: bool = Fal
                     data.append(line.strip().split())
                     break
 
-    df = pd.DataFrame(data, columns=header)
+    df = pd.DataFrame(data, columns=header)  # pyright: ignore [reportArgumentType]
     df[header[0:-2]] = df[header[0:-2]].apply(pd.to_numeric, errors="coerce")
     return df.sort_values(by=by, ascending=not reverse).to_string(index=False)
 

sonusai/utils/__init__.py

@@ -27,6 +27,7 @@ from .get_frames_per_batch import get_frames_per_batch
 from .get_label_names import get_label_names
 from .grouper import grouper
 from .human_readable_size import human_readable_size
+from .load_object import load_object
 from .max_text_width import max_text_width
 from .model_utils import import_module
 from .numeric_conversion import float_to_int16

sonusai/utils/asr_functions/aaware_whisper.py

@@ -20,7 +20,7 @@ def aaware_whisper(audio: AudioT, **_config) -> ASRResult:
 
     url = getenv("AAWARE_WHISPER_URL")
     if url is None:
-        raise EnvironmentError("AAWARE_WHISPER_URL environment variable does not exist")
+        raise OSError("AAWARE_WHISPER_URL environment variable does not exist")
     url += "/asr?task=transcribe&language=en&encode=true&output=json"
 
     with tempfile.TemporaryDirectory() as tmp:

sonusai/utils/audio_devices.py

@@ -4,14 +4,17 @@ import pyaudio
 def get_input_device_index_by_name(p: pyaudio.PyAudio, name: str | None = None) -> int:
     info = p.get_host_api_info_by_index(0)
     device_count = info.get("deviceCount")
-    for i in range(0, device_count):
-        device_info = p.get_device_info_by_host_api_device_index(0, i)
-        if name is None:
-            device_name = None
-        else:
-            device_name = device_info.get("name")
-        if name == device_name and device_info.get("maxInputChannels") > 0:
-            return i
+    if isinstance(device_count, int):
+        for i in range(0, device_count):
+            device_info = p.get_device_info_by_host_api_device_index(0, i)
+            if name is None:
+                device_name = None
+            else:
+                device_name = device_info.get("name")
+            if isinstance(device_name, str) or device_name is None:
+                input_channels = device_info.get("maxInputChannels")
+                if name == device_name and isinstance(input_channels, int) and input_channels > 0:
+                    return i
 
     raise ValueError(f"Could not find {name}")
 
@@ -20,11 +23,14 @@ def get_input_devices(p: pyaudio.PyAudio) -> list[str]:
     names = []
     info = p.get_host_api_info_by_index(0)
     device_count = info.get("deviceCount")
-    for i in range(0, device_count):
-        device_info = p.get_device_info_by_host_api_device_index(0, i)
-        device_name = device_info.get("name")
-        if device_info.get("maxInputChannels") > 0:
-            names.append(device_name)
+    if isinstance(device_count, int):
+        for i in range(0, device_count):
+            device_info = p.get_device_info_by_host_api_device_index(0, i)
+            device_name = device_info.get("name")
+            if isinstance(device_name, str):
+                input_channels = device_info.get("maxInputChannels")
+                if isinstance(input_channels, int) and input_channels > 0:
+                    names.append(device_name)
 
     return names
 
@@ -32,10 +38,13 @@ def get_input_devices(p: pyaudio.PyAudio) -> list[str]:
 def get_default_input_device(p: pyaudio.PyAudio) -> str:
     info = p.get_host_api_info_by_index(0)
     device_count = info.get("deviceCount")
-    for i in range(0, device_count):
-        device_info = p.get_device_info_by_host_api_device_index(0, i)
-        device_name = device_info.get("name")
-        if device_info.get("maxInputChannels") > 0:
-            return device_name
+    if isinstance(device_count, int):
+        for i in range(0, device_count):
+            device_info = p.get_device_info_by_host_api_device_index(0, i)
+            device_name = device_info.get("name")
+            if isinstance(device_name, str):
+                input_channels = device_info.get("maxInputChannels")
+                if isinstance(input_channels, int) and input_channels > 0:
+                    return device_name
 
     raise ValueError("No input audio devices found")

sonusai/utils/docstring.py

@@ -1,4 +1,4 @@
-def trim_docstring(docstring: str) -> str:
+def trim_docstring(docstring: str | None) -> str:
     """Trim whitespace from docstring"""
     from sys import maxsize
 
@@ -30,11 +30,14 @@ def trim_docstring(docstring: str) -> str:
     return "\n".join(trimmed)
 
 
-def add_commands_to_docstring(docstring: str, plugin_docstrings: list[str]) -> str:
+def add_commands_to_docstring(docstring: str | None, plugin_docstrings: list[str]) -> str:
     """Add commands to docstring"""
     import sonusai
 
-    lines = docstring.splitlines()
+    if not docstring:
+        lines = []
+    else:
+        lines = docstring.splitlines()
 
     start = lines.index("The sonusai commands are:")
     end = lines.index("", start)

sonusai/utils/energy_f.py

@@ -29,14 +29,16 @@ def compute_energy_f(
         if transform is None:
             raise ValueError("Must provide ForwardTransform object")
 
-        frequency_domain = transform.execute_all(torch.from_numpy(time_domain))[0].numpy()
+        _frequency_domain = transform.execute_all(torch.from_numpy(time_domain))[0].numpy()
+    else:
+        _frequency_domain = frequency_domain
 
-    frames, bins = frequency_domain.shape
+    frames, bins = _frequency_domain.shape
     result = np.empty((frames, bins), dtype=np.float32)
 
     for f in range(frames):
         for b in range(bins):
-            value = frequency_domain[f, b]
+            value = _frequency_domain[f, b]
             result[f, b] = np.real(value) * np.real(value) + np.imag(value) * np.imag(value)
 
     return result

sonusai/utils/human_readable_size.py

@@ -1,7 +1,7 @@
-def human_readable_size(size: float, decimal_places: int = 3) -> str:
+def human_readable_size(num: float, decimal_places: int = 3, suffix: str = "B") -> str:
     """Convert number into string with units"""
-    for unit in ["B", "kB", "MB", "GB", "TB"]:  # noqa: B007
-        if size < 1024.0:
-            break
-        size /= 1024.0
-    return f"{size:.{decimal_places}f} {unit}"
+    for unit in ("", "k", "M", "G", "T", "P", "E", "Z"):
+        if abs(num) < 1024.0:
+            return f"{num:.{decimal_places}f} {unit}{suffix}"
+        num /= 1024.0
+    return f"{num:.{decimal_places}f} Y{suffix}"

sonusai/utils/load_object.py

@@ -0,0 +1,15 @@
+from functools import lru_cache
+from typing import Any
+
+
+@lru_cache
+def load_object(name: str) -> Any:
+    """Load an object from a pickle file (with LRU caching)"""
+    import pickle
+    from os.path import exists
+
+    if exists(name):
+        with open(name, "rb") as f:
+            return pickle.load(f)  # noqa: S301
+
+    raise FileNotFoundError(name)

sonusai/utils/onnx_utils.py

@@ -3,8 +3,8 @@ from collections.abc import Sequence
 from onnx import ModelProto
 from onnx import ValueInfoProto
 from onnxruntime import InferenceSession
-from onnxruntime import NodeArg
-from onnxruntime import SessionOptions
+from onnxruntime import NodeArg  # pyright: ignore [reportAttributeAccessIssue]
+from onnxruntime import SessionOptions  # pyright: ignore [reportAttributeAccessIssue]
 
 REQUIRED_HPARAMS = (
     "feature",

sonusai/utils/print_mixture_details.py

@@ -37,10 +37,10 @@ def print_mixture_details(
         print_fn(f'{"  Samples":{desc_len}} {mixture.samples}')
         print_fn(f'{"  Feature frames":{desc_len}} {mixdb.mixture_feature_frames(mixid)}')
         print_fn(f'{"  Noise file":{desc_len}} {noise_file.name}')
-        noise_offset_percent = int(np.round(100 * mixture.noise.offset / float(noise_file.duration * SAMPLE_RATE)))
-        print_fn(f'{"  Noise offset":{desc_len}} {mixture.noise.offset} samples ({noise_offset_percent}%)')
+        noise_offset_percent = int(np.round(100 * mixture.noise_offset / float(noise_file.duration * SAMPLE_RATE)))
+        print_fn(f'{"  Noise offset":{desc_len}} {mixture.noise_offset} samples ({noise_offset_percent}%)')
         print_fn(f'{"  SNR":{desc_len}} {mixture.snr} dB{" (random)" if mixture.snr.is_random else ""}')
-        print_fn(f'{"  Target gain":{desc_len}} {[target.gain for target in mixture.targets]}')
+        print_fn(f'{"  Target gain":{desc_len}} {[target.gain if not mixture.is_noise_only else 0 for target in mixture.targets]}')
         print_fn(f'{"  Target SNR gain":{desc_len}} {mixture.target_snr_gain}')
         print_fn(f'{"  Noise SNR gain":{desc_len}} {mixture.noise_snr_gain}')
         print_fn("")

{sonusai-0.19.6.dist-info → sonusai-0.19.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sonusai
-Version: 0.19.6
+Version: 0.19.9
 Summary: Framework for building deep neural network models for sound, speech, and voice AI
 Home-page: https://aaware.com
 License: GPL-3.0-only
@@ -27,7 +27,7 @@ Requires-Dist: pandas (>=2.2.3,<3.0.0)
 Requires-Dist: pesq (>=0.0.4,<0.0.5)
 Requires-Dist: praatio (>=6.2.0,<7.0.0)
 Requires-Dist: psutil (>=6.0.0,<7.0.0)
-Requires-Dist: pyaaware (>=1.5.18,<2.0.0)
+Requires-Dist: pyaaware (>=1.6.3,<2.0.0)
 Requires-Dist: pyaudio (>=0.2.14,<0.3.0)
 Requires-Dist: pydub (>=0.25.1,<0.26.0)
 Requires-Dist: pystoi (>=0.4.1,<0.5.0)

{sonusai-0.19.6.dist-info → sonusai-0.19.9.dist-info}/RECORD

@@ -1,7 +1,7 @@
-sonusai/__init__.py,sha256=AzY-ysnmnzg0dDsephK12AuPoClcf8ggphjk4i9Qe-I,2762
-sonusai/aawscd_probwrite.py,sha256=DZwvkrNy-v4meCShSNAEtlwLnd-3OcpbHkHYvJt2jFI,3652
+sonusai/__init__.py,sha256=NSb0bvmAh6Rm2MDtchpAGsg8a3BrmVnShYb-vC_emH8,2802
+sonusai/aawscd_probwrite.py,sha256=QZLMQrmPr3OjZ06buyYDwlnk9YPCpyr4KHkBjPsiqjU,3700
 sonusai/audiofe.py,sha256=iFdthh4UrOvziT8urjrjD7dACWZPQz9orM5bVAW3WSQ,11269
-sonusai/calc_metric_spenh.py,sha256=w7lCuAI01B_E_pWQnGlGQQcmzopi7q3QeO5ByMXnzos,46712
+sonusai/calc_metric_spenh.py,sha256=NTqYBNDR6LTSdDA8s4WCSEwajE3-fNJg1TUzhNzJC9g,46719
 sonusai/data/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 sonusai/data/genmixdb.yml,sha256=U_kLbE7gZ5rA7yNSB2NW7eK5dnYP5grJVMR321VMLt8,940
 sonusai/data/speech_ma01_01.wav,sha256=PK0vMKg-NR6rPE3KouxHGF6PKXnJCr7AwjMqfu98LUA,76644
@@ -12,68 +12,69 @@ sonusai/deprecated/tplot.py,sha256=0p238DvTaP4oU9y-dp0JdLaTV4TKrooAwbx7zdz_QAc,1
 sonusai/doc/__init__.py,sha256=KyQ26Um0RM8A3GYsb_tbFH64RwpoAw6lja2f_moUWas,33
 sonusai/doc/doc.py,sha256=VZXauwbOb-VIufWw-lu0yfrd6jMRPeFeVPaaEjZNvn4,18881
 sonusai/doc.py,sha256=zSmXpioB0YS_5-7kqfS5cr--veSaXkxRKzldId9Hyoc,878
-sonusai/genft.py,sha256=z9J4-aYZdPw50_56QMOSaofLNadO5b7htxtDgd0Qu88,5227
-sonusai/genmetrics.py,sha256=G1lIgeIAu5EmalDR7j0rbSRP0dyndY4w35xUT5iToYU,4958
-sonusai/genmix.py,sha256=4YEPvUnFCAexwW540LyzXsFq9QODE4GgOrCZOvigVJA,6179
-sonusai/genmixdb.py,sha256=UmkiQTK8W5hRBVGlCBCJnBBnXCK1yJj7bhuXNtyGM38,19203
+sonusai/genft.py,sha256=TqtmexKw7tZsugU1DbZ3fY7_YZ2hFlIU-ema6f0LRgQ,5586
+sonusai/genmetrics.py,sha256=RiEYmkRl0yVCIr9PHkKRqEG68WR77pP9WpWAYeKnMLE,5723
+sonusai/genmix.py,sha256=pugp74prksf4ZiJYMXWg2l3v48Qqi9KpP2WhPEBpZOU,6702
+sonusai/genmixdb.py,sha256=gUF_9dGfuWRCmHtCoXZRrgCMejz46A5KhWoEh31laqc,18724
 sonusai/lsdb.py,sha256=0HOGDDndB3LT9cz9AaxKIpt9vslAoSP4F239gply4Xg,5149
 sonusai/main.py,sha256=HbnEia1B1-Z-mlHkLfojH8aj9GIpL1Btw3oH60T_CCQ,2590
 sonusai/metrics/__init__.py,sha256=ssV6JEK_oklRSocsp6HMcG-GtJvV8IkRQtdKhHHmwU8,878
 sonusai/metrics/calc_audio_stats.py,sha256=IHgYEPaob_Nw35SaH3tyHp7Wwju4f-2-BJZ99JyeLmc,1572
-sonusai/metrics/calc_class_weights.py,sha256=DP36f0W5Vr8v8LhzmOWFxb1cgXFV-pzN6i4lbG7xMhw,3796
-sonusai/metrics/calc_optimal_thresholds.py,sha256=y2bTF409iJBOK0WblJ62XcF83BT3QvPuu_nb7xH4Vt8,3433
+sonusai/metrics/calc_class_weights.py,sha256=SUOCdM4w03rFpyxAriPnPwCtEEFsAH3WxpK9N_fupwo,3637
+sonusai/metrics/calc_optimal_thresholds.py,sha256=0JOqU__doeOpNtgEZgeO1Kg7pttJRpITTVmqLU6TadY,3513
 sonusai/metrics/calc_pcm.py,sha256=yBQV9UJ1GK5f4c_8TNABMtZR-xyStKJCsSTT0FQGa50,1886
 sonusai/metrics/calc_pesq.py,sha256=dCztUTaPyyjkUI2DpejqhiPzQv4FOtigzffFnDXDs-M,949
-sonusai/metrics/calc_phase_distance.py,sha256=-aNZEkfqxv3UjQsYcGqJ-Ips-ZAktBcykGJ6O2seHjA,1863
+sonusai/metrics/calc_phase_distance.py,sha256=MFuBtGXb5qfQvRVciJ0Soz1w0GMSeJLBJud-aK4Loow,1870
 sonusai/metrics/calc_sa_sdr.py,sha256=IdCzlQ_w94A3eK42t-gy_DrFN_tziwfDjTU6-WKuCFs,2531
 sonusai/metrics/calc_sample_weights.py,sha256=0O2EH1-FKlCa0HFgKPUF1BJiknR1hCH7zLbXnoXH7Ag,945
 sonusai/metrics/calc_segsnr_f.py,sha256=-ncM0OGRcOBDh2PnGsQOQccHs1QXHBTd3eCv4YpF3fs,2923
-sonusai/metrics/calc_speech.py,sha256=8QhOpupbQOte9l3a4RsUrmSsG0qXtAeMcGfT2tMDzhs,14776
+sonusai/metrics/calc_speech.py,sha256=3Gs6lmEaZj7XPZYuQnal2_Wa-z-6naktSfLRnl0H53g,14754
 sonusai/metrics/calc_wer.py,sha256=1MQYMx8ldHeodtJEtGibvDKhvSaGe6DBmZV4L8qOMgg,2362
 sonusai/metrics/calc_wsdr.py,sha256=vcALY-zuhyThRa1QMz2qW8L9kSBc2v32gV9u8bV7VaM,2556
-sonusai/metrics/class_summary.py,sha256=qBWxtQ9YKJUzk5LsjDuwnYEKrpfwubF-vZZL0HbcSH8,3193
-sonusai/metrics/confusion_matrix_summary.py,sha256=eZFmt7k5Eanx7FYPwcfnzVq-AMYNSP03gfj9GtsfHWI,3841
-sonusai/metrics/one_hot.py,sha256=jo-YsiDVr3PYoZTHhA0t24-VpSXEsZ4lukrP1xFbXAQ,13810
-sonusai/metrics/snr_summary.py,sha256=sRUE6XI5PFFhaSpag537WUnukbMizOeMCwVQ2ww34Cw,5617
-sonusai/mixture/__init__.py,sha256=NeXnrPIgDB-33WA0wva7m41SGOiO6biyKiDKK9o_W28,5650
+sonusai/metrics/class_summary.py,sha256=ZA7zNgwBpmTs1TP_t4jRT0pWnDnATC_up_8qE4aH-do,2809
+sonusai/metrics/confusion_matrix_summary.py,sha256=zBL_Ke7wF6oKtrKZPr0fsyF_taofdjxBlZmKodu0xUA,3143
+sonusai/metrics/one_hot.py,sha256=hmuyh-9tpRjb_oyqU3WqZ14zItpRJQfcqBDKJeb5H9I,13930
+sonusai/metrics/snr_summary.py,sha256=t8Fi_8WtboTi8flkZuOiHq9H3-nIELx4AKvnm-qvxLQ,5785
+sonusai/mixture/__init__.py,sha256=ccz_UpB7Rh6wUB7uwLgNhciMZhKE9Qt9ICLyjq08kxs,5126
 sonusai/mixture/audio.py,sha256=Tg0-HzSgZDttKwWwONqgK6qK4maTSmWcfiL1atZcIXQ,2390
-sonusai/mixture/augmentation.py,sha256=QqAuM2mN5S4SkQtCg1eDSJEtGsxsa6MBYZ9IOGVBvxg,10372
-sonusai/mixture/class_count.py,sha256=U_FiWBU_yH6_ZAmZRdeDKUQyS0fnTtyfJ2V1lgqkFls,605
-sonusai/mixture/config.py,sha256=b7fyacZm2FVjcTYC1vOy5FLrxkuJRYR0KrrATcGwQK4,24658
+sonusai/mixture/augmentation.py,sha256=s8QlPHnFJOblRU59fMQ-Zqysiv4OUJ7CxLRcV81lnaA,10407
+sonusai/mixture/class_count.py,sha256=zcC3BDYMPN6wJYmO1RcOuqmrnTQIbMSznl33oN3e2sc,597
+sonusai/mixture/config.py,sha256=g5ZmOhFYqmEdRQYSgfDIZ9VM0QiTwBqk7vIyAvxnPMo,24211
 sonusai/mixture/constants.py,sha256=fXcWuSI4YZOAuncLGEUeEW9WWNZeN-6mI8LFNILwyTc,1494
-sonusai/mixture/data_io.py,sha256=s8DNUIwdgcgrpqP_JplgGB_zVBczgGVTtAZe2qqrtiI,4632
-sonusai/mixture/datatypes.py,sha256=D20qe93HjXZ80R_lNgN3Pu4pZH2qTxf2ZmntknDGNrs,10355
-sonusai/mixture/db_datatypes.py,sha256=kQG_-G3zOqixoHxcmTrZlmxdrWNr1VhUzYF5IKYi19c,1542
+sonusai/mixture/data_io.py,sha256=KZGqhHd9_ucAfZEAXPIc5XL3aHYgdV5CyqaBx5_t8OM,5551
+sonusai/mixture/datatypes.py,sha256=xNDBWFTVQ3plJ7qHKzrXyV4pffPYuf1xMVqBsR40n4o,10487
+sonusai/mixture/db_datatypes.py,sha256=kvdUOMS6Pkkj9AmxCiq6zM8x7jbPPi933tVaXRxbTdQ,1534
 sonusai/mixture/eq_rule_is_valid.py,sha256=O3gCAs_0hpxENK5b7kxxpDmOpKHlXGBWuLGT_97ARSM,1210
-sonusai/mixture/feature.py,sha256=q9QZRe1qfKvHOuDxLAbpdrQF1gIqL8NRvk4WYWZX2IU,2996
-sonusai/mixture/generation.py,sha256=SMRYhTYmuUHejR894HDGdYSDOZvEdFobdfQPCTopB-E,40218
-sonusai/mixture/helpers.py,sha256=1La31bTdXA5gFCHRqQWDk6vfNlpKG1AX2jqlhOnObqU,20834
+sonusai/mixture/feature.py,sha256=L0bPFG0RO-CrrtTStUMt_14euYsVo8_TWTP2IKSFKaA,2335
+sonusai/mixture/generation.py,sha256=Okmyc7LVVdyt7UDzLSR4XdK-Q92vSQHX0RI__RJbJM8,38551
+sonusai/mixture/helpers.py,sha256=UEN_9LttIZwvms9nOkjMr_Yf3lpb7sMPmDVUoX8D7yQ,15225
 sonusai/mixture/log_duration_and_sizes.py,sha256=qhgl87C2KbjxLdKEpjYOoqNL6rc-8-PB4R7Gx_7UG8g,1240
-sonusai/mixture/mixdb.py,sha256=Qj_12vZ4_4CdKrkWdBa0ysBt9LSCgs6YdJoiH3L9YQ0,65790
-sonusai/mixture/soundfile_audio.py,sha256=fAbNIW8Xi7tfglmrbOzMSLQIG0dZEqTjLxhPGUeFyFc,4073
-sonusai/mixture/sox_audio.py,sha256=QnHToWYy1l9kKGIefciUf6CdcAphDiaWcHo4imukUgY,16522
-sonusai/mixture/sox_augmentation.py,sha256=txaoSdbLdizrzBbdV0LYleM7lxitV3h27_C68Fij1GY,4506
+sonusai/mixture/mixdb.py,sha256=W8B54U6f-8szMylXSnU0YzOR-a9M3XoCRpyqOgBIJuQ,73498
+sonusai/mixture/soundfile_audio.py,sha256=At_ZC2b9pZ_9IYp1UxyPzRoBK9-1cKPCLMm74F1AjKE,4092
+sonusai/mixture/sox_audio.py,sha256=7ouCLqXYS6tjG2L0v5lugVO7z5UwJmsr1VigbrXhs74,16725
+sonusai/mixture/sox_augmentation.py,sha256=DtfGLPaB1BIt2wvTEA__MYkGFNU85Tuup5BFsIVrh0E,4546
 sonusai/mixture/spectral_mask.py,sha256=U9XJ_SAoI9b67K_3SE7bNw6U8cPGFOBttaZAxMjA_Jc,2042
-sonusai/mixture/target_class_balancing.py,sha256=B8h9fqdvzZgcpsi-Z91GBEi4YSZJj0uMGzNCeGQk7RU,4313
-sonusai/mixture/targets.py,sha256=QmhidE9wWU4oeumtUTHrndlcl0cG8_TiIzltnPwfBOg,6397
+sonusai/mixture/target_class_balancing.py,sha256=o_TZ8kVYq10lgeXHh3GUFfflfdUvRt0FekFu2eaNkDs,4251
+sonusai/mixture/targets.py,sha256=6emo2fxxp9ZhSpHuUM9xIjYMz8zeIHAw684jT3l7fAs,6442
 sonusai/mixture/tokenized_shell_vars.py,sha256=lXTzUDutuBWGV1zIsqeIxWmy-eKm0Vx1y8-iLdsL1gQ,4921
-sonusai/mixture/torchaudio_audio.py,sha256=169VXKEoOHc5nyiHJwaihkcQ_a1ZH_O-nnk9Gq4JtaQ,2887
-sonusai/mixture/torchaudio_augmentation.py,sha256=68QpJ4JcAH2fsL4qVck7VvBv615UbCBPEnNe4HMvA_8,3844
-sonusai/mixture/truth.py,sha256=IJ9BHrBOMlFS8UGlxuw0-Uao0kHX4mDCmcKXxDs0Lzs,2342
-sonusai/mixture/truth_functions/__init__.py,sha256=MEjNPOCd-cXLHpN09GuMv_DPIVWLLI6uhxsiKoaj6JM,1285
-sonusai/mixture/truth_functions/crm.py,sha256=mE0q7ax37ZUMVFRi5EW3uqFB8v50WkLB4tgPYdCPjAc,3212
-sonusai/mixture/truth_functions/datatypes.py,sha256=fbh16Q4Du4uCpQE_VGp8r3rhZTaRGK3glkkDmJ9BqPY,1285
-sonusai/mixture/truth_functions/energy.py,sha256=4uYBEaU-lPnmzJGROa77qWizDMNW2M-c7imOqApmXJE,5451
-sonusai/mixture/truth_functions/file.py,sha256=1jw7C3d45NXh4AZYfycyZm1aipX4xgEkpS-GZGgdJ6Y,1339
-sonusai/mixture/truth_functions/phoneme.py,sha256=5DbELzq6CDKmjvVndQzq_RzqOLCL7oBqiMEsAuak_UU,855
-sonusai/mixture/truth_functions/sed.py,sha256=2Aq_q5POrGbfpBCPFbuXY7LTjiVi5-gNHHr6-mezZqs,3471
-sonusai/mixture/truth_functions/target.py,sha256=FaIlsQUqEpP7pokCVQ_x19SLKzRT0Ncw4rgbv0e4_dw,3664
-sonusai/mkwav.py,sha256=scSULegmcm7zCCbi4YVI5sIEumxy2T01yNOQt5DlmOI,4064
+sonusai/mixture/torchaudio_audio.py,sha256=72Hxo5TKAW7mYpRy15QFfD7AYDORBk6bVCcHENniWGw,3116
+sonusai/mixture/torchaudio_augmentation.py,sha256=uFAKxIfs50J5FR-WXodsEACm2Ao-t5dZRSJ0DwTAfBg,3930
+sonusai/mixture/truth.py,sha256=-CwwawFRGjqodR2yKvAMGL1XaYLct-tli7wZ2gbhLtQ,2121
+sonusai/mixture/truth_functions/__init__.py,sha256=0mlOFChPnXG5BC0eKOe4n9VH17jY4iOqZFLuF6Gprdk,1505
+sonusai/mixture/truth_functions/crm.py,sha256=iidcffXfqV8k9O5wt5KTWIAFaTSjmhV5ucKZPbTgpvQ,3809
+sonusai/mixture/truth_functions/energy.py,sha256=dlxl5c58bv012KgyKeO78urXSh6FK3LdqiKk0SDkmS8,6540
+sonusai/mixture/truth_functions/file.py,sha256=pyCAhx3PhJRBoZMrjoQI4Tbi5TN7sPembSVEr80Bu3g,1431
+sonusai/mixture/truth_functions/metadata.py,sha256=aEZly5bJEaZpUBZonWvcu14_Dn3M2HamwTaM5Bg7Tm8,778
+sonusai/mixture/truth_functions/metrics.py,sha256=xfH98eXBe20kwRk-tdWbBkOplF1cCA-u_naJUiTps0Y,873
+sonusai/mixture/truth_functions/phoneme.py,sha256=jwBYiNwwBwh2tHtOJ2NopYWhT6y19kXzSIag0XW9GSY,778
+sonusai/mixture/truth_functions/sed.py,sha256=C0n9DkfBNQblFsFCkPbooy54KuHSY7B0f1vLft2asdw,3832
+sonusai/mixture/truth_functions/target.py,sha256=nSkHFESzCEOljcYf4jQ7FmxsAWJtMCRRWFKM_DyjoLU,4926
+sonusai/mkwav.py,sha256=xoqrX76gC1zE-Xb1Zfm9ZpcZ95l0qF1ZyOPmLMkFomk,4053
 sonusai/onnx_predict.py,sha256=Y1VUN0wuvloEW46uxg7X4ywaec_Xx92djCU3BP0KAx0,8699
 sonusai/queries/__init__.py,sha256=bhoeOFfu9GA5DOUuxRrIev7MYdXaGN8xdKJ6BXyNNtQ,277
-sonusai/queries/queries.py,sha256=u8tZCJVnw9wJlwc03ue63FFqD5CHAKe4PWLJk0z4rto,7589
+sonusai/queries/queries.py,sha256=N84UcRrgDPLGJZesS2KQ9eyRzPRSyukjOlc786h5Ehs,7509
 sonusai/speech/__init__.py,sha256=vqAymCBPjMUSM4OZKHTai6BYwXsOBlf_G_vOhELVf8I,133
-sonusai/speech/l2arctic.py,sha256=Gon9OPAtcI0kItQ87wuyrvNqDpo_Lad3N2lOXyLSwkY,3631
+sonusai/speech/l2arctic.py,sha256=VQNKuTbmlbW0PJ7bOjx9sr0VjUYxJnxfTiPJIa4OOaA,3829
 sonusai/speech/librispeech.py,sha256=ugP3NVOenSsBF1cUG4Nyl7dumGHQmE4Ugk1yYjtOyj4,3070
 sonusai/speech/mcgill.py,sha256=sgPHEZTPHlFXF8GVYFfKXMUEyBikfKha2RWOPfpNy_U,1981
 sonusai/speech/textgrid.py,sha256=WvsUeamKoYyXBNnNnZgug-xfTiq2Z6RnFc1u0swmqNw,2813
@@ -81,13 +82,13 @@ sonusai/speech/timit.py,sha256=B1DZCS5Crt9Y54drqVqurhEiOMUZGxVOW7gxzZA5ErY,4099
 sonusai/speech/types.py,sha256=4eKVPAktpkIrZ2qoVp2iT45zxTVNocQEGT6O_Zlub_w,214
 sonusai/speech/vctk.py,sha256=WInvRRRkZCW6t_NcZAJffJzgCbyetal-j2w0kKX5SDw,1527
 sonusai/speech/voxceleb.py,sha256=Uu1kB1krf8hess1yuvGbYfV_VgYhklEyoz4I7KfrVpw,2658
-sonusai/summarize_metric_spenh.py,sha256=2HBFfVlezIoz18DTBi5rLyUqWy4QL87OXahMaJ6NDUQ,1818
-sonusai/utils/__init__.py,sha256=BAo0J8ZqDHiTb6dpfBu0KrueCtsSlrtOZ9Kstst3kBk,2342
+sonusai/summarize_metric_spenh.py,sha256=2w81ZgJahYvD6wCpE3DFoUFrXexLXjO44ITRVm1HJXw,1858
+sonusai/utils/__init__.py,sha256=z72OlzZCHpYfYHKnHn7jznj6Zt7zB-FyO6hIgFk45As,2379
 sonusai/utils/asl_p56.py,sha256=cPUVwXawF7vLJgs4zUtoRGk7Wdbe5KKti_-v_8xIU10,3862
 sonusai/utils/asr.py,sha256=20eKkatNqGrbxFQoyeCRzqXaSH3lcA-9ZXGneCXjLvs,2791
 sonusai/utils/asr_functions/__init__.py,sha256=HKGRm_c48tcxlfwqH63m-MvhAoK_pCcw76lxmFmiP_U,63
-sonusai/utils/asr_functions/aaware_whisper.py,sha256=MHH238wj6arkQUMDHqHjc4DDVY2sXzu6iGi3ISswTeM,2129
-sonusai/utils/audio_devices.py,sha256=Z_xstSy8KeJ2xtzHJGmrIHyCiNMgpdpJd5T-19u8dGU,1421
+sonusai/utils/asr_functions/aaware_whisper.py,sha256=M9Y8Pgh1oIrDOPZZPSRPDig8foxfgs3f8AsoZ8W00B0,2120
+sonusai/utils/audio_devices.py,sha256=_Eiah86SZjbdp2baD2AUVF4FmhseiNuG3KJkd_LbULk,2041
 sonusai/utils/braced_glob.py,sha256=uvxo7USbxH6cWuVdNeGrz1SbZISFR1gPGbpy0EWm3m8,1645
 sonusai/utils/calculate_input_shape.py,sha256=TIa_rHW3VIvOhlv5Wa3orcWFPMT-a3EPrIFHstbrmo4,906
 sonusai/utils/compress.py,sha256=tT983XlgRf9bzutCegtHzkfKHK5LZD6a_mMFP47FGMI,605
@@ -96,20 +97,21 @@ sonusai/utils/create_timestamp.py,sha256=s7ALOX3MAyK3EOX2BVOiYTIcspsKlIM6zXJk2cb
 sonusai/utils/create_ts_name.py,sha256=3xu10hbZkV18sDqK4oR1JYvXeYE53ufzddmvGYx83Vg,405
 sonusai/utils/dataclass_from_dict.py,sha256=iUagjF7CzbDIBKTX4ktd7EXn8q1jxmiRClMH3fu2_oA,389
 sonusai/utils/db.py,sha256=lI77MJJLs4CTYxhjFUvBom2Kk2imAP34okOeO4irbDc,371
-sonusai/utils/docstring.py,sha256=nJEwaVlOuSEZiSH34zwNdfGNxrY0NLCTZ7VoLzc9-cU,1447
-sonusai/utils/energy_f.py,sha256=4gN5mNGuHgCgUbnOQ9ReYkDqNvE2pueoYBYWE-GBQPU,1457
+sonusai/utils/docstring.py,sha256=aPFrVpqlbYq3kIJE_sLCjhXWEqWCBo_q_DUmEXcGONQ,1516
+sonusai/utils/energy_f.py,sha256=k1S5ELsNsm3Sn0RogFEzZv1bW-X8yl2dc2KjWne_t2I,1515
 sonusai/utils/engineering_number.py,sha256=SToFf6S0Xu0NtAJ1SjsVH4wxylH7qK8S9TBkPa15opY,5510
 sonusai/utils/get_frames_per_batch.py,sha256=xnq4tV7MT74N0H6b5ZsiAezqdXucboCLQw1Np9XpZbs,134
 sonusai/utils/get_label_names.py,sha256=df4jZVaQ3WnYQqNj21iUV4aYWyQEZUNmgs93qKW-_rA,820
 sonusai/utils/grouper.py,sha256=qyZ0nj84yOrC-RZsXHC-KJvcUliGktnV8S6-P3PD6_w,203
-sonusai/utils/human_readable_size.py,sha256=CmLxq9y5f-OPkwTEu4Gm0Io-H6p4SRhD2ciJRuYXfWk,293
+sonusai/utils/human_readable_size.py,sha256=DOCS7SAymrtTZli8AczvyCMCh44r7ZDgVBA7jSZupmA,356
+sonusai/utils/load_object.py,sha256=-XMMwSNUPhtg6G6FhLwL1zvxwu-PLwj25QYUEielW2k,356
 sonusai/utils/max_text_width.py,sha256=pxiJMwb_zlkNntexgo7S6lAuF7NLLZvFdOCkxdsQJVY,315
 sonusai/utils/model_utils.py,sha256=OIJBhOjxR0wpxsd7A2r6J2AjqfdYgZzi6UEThw4S1lI,828
 sonusai/utils/numeric_conversion.py,sha256=iFPXFU8C_1mW5tmDqHq8-xP1tL8nVaSmhQRakdCqy30,328
-sonusai/utils/onnx_utils.py,sha256=mbat4xvLhKMSvMC2kTEc4ukc7PI_gjsG4IVGvzpcOd4,5764
+sonusai/utils/onnx_utils.py,sha256=cN8NZDrERPvbbJFSGI9YzseCezAz-Bckfel-EhkfRUU,5860
 sonusai/utils/parallel.py,sha256=yvRZvZWPR5slM51i08m7sYx-Mvsb5oryCqqJXVoJ8tQ,2190
 sonusai/utils/path_info.py,sha256=QY7iQ0nYpeEDnPN9RyPh4DsgYmVYsLrrlAzKuzkqX1o,118
-sonusai/utils/print_mixture_details.py,sha256=Wh5M0aTy3NzxvR3P0HDlSq_S3G9L7LoQx-FnDCrHODs,2911
+sonusai/utils/print_mixture_details.py,sha256=EypVeiQ8IjA0rGVfU91S3IIhzgGjkhU2oJ0DKJUBbWQ,2947
 sonusai/utils/ranges.py,sha256=-TtAR0Vg_j4kYtJOvEOYQllBZEat_KfUKsfRxr5oj-o,1235
 sonusai/utils/read_predict_data.py,sha256=PUSroxmWQGtr6_EcdSHmIFQoRGou8CKKqcggWylfTqQ,1056
 sonusai/utils/reshape.py,sha256=Ozuh3UlmAS5NCeOK7NR8KgcQacHvgq10pys0VfCnOPU,5746
@@ -119,7 +121,7 @@ sonusai/utils/stratified_shuffle_split.py,sha256=d7WLUirywSvgZWkt_5a0F8YvTnJjuXl
 sonusai/utils/write_audio.py,sha256=0lKdaX57N6H-UWdioqmXCJMjwT1eBz5B-bSGqDvloAc,838
 sonusai/utils/yes_or_no.py,sha256=0h1okjXmDNbJp7rZJFR2V-HFU1GJDm3YFTUVmYExkOU,263
 sonusai/vars.py,sha256=kBBzuvC8szmdIZEEDA7XXmD765addZKdM2aFipeGO1w,933
-sonusai-0.19.6.dist-info/METADATA,sha256=29GCEKJsRdTRsDJw4_wdxUWgKckdPBDPO2WVo9Bz_sY,2536
-sonusai-0.19.6.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-sonusai-0.19.6.dist-info/entry_points.txt,sha256=zMNjEphEPO6B3cD1GNpit7z-yA9tUU5-j3W2v-UWstU,92
-sonusai-0.19.6.dist-info/RECORD,,
+sonusai-0.19.9.dist-info/METADATA,sha256=rCXEzWOsVKZdZB8E5iPt0fachIQ_-zkmM1UNAe8JZc4,2535
+sonusai-0.19.9.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+sonusai-0.19.9.dist-info/entry_points.txt,sha256=zMNjEphEPO6B3cD1GNpit7z-yA9tUU5-j3W2v-UWstU,92
+sonusai-0.19.9.dist-info/RECORD,,