API 參考

這一頁的每個名稱都直接來自 import kudio：呼叫的是 kudio.file_load，不是 kudio.core.io.file_load。後面幾頁依模組記載同樣這些名稱，附上完整的簽名、每個參數，以及 docstring 裡的範例。API 的說明取自程式碼裡的 docstring，所以維持英文。

音訊檔案

完整說明在音訊檔案。

名稱	種類	用途（取自 docstring）
audio_info	函式	Describe an audio file (or an in-memory waveform) without interpreting it.
AudioDataManager	類別	Keyed store of audio clips: {name: {dir, wave, sr, ch}}.
AudioInfo	類別	What a file is, before you decide whether to load it.
check_file	函式	Normalize an output wave path: force the .wav suffix, create parent directories, and (optionally) rename to avoid overwriting an existing file.
check_input	函式	Recursively collect .wav files from input.
check_path	函式	Return path if it is free (or exist_ok), otherwise an incremented variant, e.g.
convert_folder	函式	Resample / re-encode / trim / normalise every .wav file under src.
ConvertResult	類別	What convert_folder did.
copy_waves	函式	Copy wave files into directory path and return where they went.
file_load	函式	Load a wave file;
load_wave	函式	Load one wave file or a list of them into (waveform, sr) tuples.
load_waves	函式	Convert every .wav in wave (see check_input) to waveform data.
LoadAudio	類別	Iterate over the .wav files found (recursively) in a directory, collected the way check_input collects them.
resample	函式	Resample a waveform already in memory.
save_wave	函式	Write wave to an audio file via soundfile.

STFT 與特徵

完整說明在STFT 與特徵。

名稱	種類	用途（取自 docstring）
ContextFrames	類別	stack_context over a whole corpus, stacked a batch at a time.
denoise_spec2wav	函式	Inverse of denoise_wav2spec;
denoise_wav2spec	函式	Peak-normalized magnitude spectrogram + phase, for denoising models.
features2matrix	函式	Concatenate feature matrices;
file_to_spectrogram	函式	File -> log-power spectrogram (see waveform_to_spectrogram).
frame_windows	函式	Cut a feature matrix into fixed-length windows.
logspec_from_files	函式	Convert each file to a log spectrogram and stack them.
melspectrogram	函式	Audio -> stacked log-mel feature vectors (n_vectors, n_mels * n_frames).
mfcc	函式	Waveform -> MFCC, shaped (1, n_frames, n_mfcc).
mfcc_from_files	函式	Load each wave file and stack their MFCC matrices (n_files, frames, n_mfcc).
save_spectrogram_as_wave	函式	Reconstruct a waveform from enhanced_spec (phase from noisy_file) and write it to wave_out_dir as 16-bit PCM, clipped at full scale.
spectrogram_to_waveform	函式	Log-power spectrogram -> waveform, reusing the phase of reference y.
SpectrogramStream	類別	A rolling log-magnitude spectrogram, one column per hop.
stack_context	函式	Stack ±context neighbouring frames onto each frame.
Standardizer	類別	Per-column mean/std normalisation, with the statistics saved alongside.
STFT	類別	Immutable STFT settings, with the two transforms attached.
wave_separate	函式	Split an interleaved multi-channel waveform into per-channel arrays.
wave_slicing	函式	Slice a waveform into overlapping frames (flattened if is_reshape).
waveform_to_spectrogram	函式	Waveform -> log-power spectrogram (frames, bins), log10(\|X\|^2).

檢查

完整說明在檢查。

名稱	種類	用途（取自 docstring）
audio_report	函式	Inspect y without needing a clean reference to compare it against.
AudioReport	類別	Reference-free measurements of one clip.
dnsmos	函式	Predict P.835 opinion scores for y.
DnsmosScore	類別	Three P.835 opinion scores, each on the 1–5 MOS scale.
find_dnsmos_model	函式	Locate sig_bak_ovr.onnx.
loudness	函式	Gated integrated loudness of y, in LUFS.
loudness_over_time	函式	Loudness measured over a sliding window -- where a take drifted.
loudness_range	函式	Loudness range in LU, to EBU Tech 3342.
LoudnessCurve	類別	Loudness measured over a sliding window, in LUFS.
match_loudness	函式	Scale y until it is as loud as reference.
MOMENTARY	常數	Convert a string or number to a floating-point number, if possible.
normalize_lufs	函式	Scale y to lufs integrated loudness.
SHORT_TERM	常數	Convert a string or number to a floating-point number, if possible.
true_peak	函式	Peak of the reconstructed waveform, in dBTP (BS.1770-4 Annex 2).

語音活動與音高

完整說明在語音活動與音高。

名稱	種類	用途（取自 docstring）
f0	函式	Track the fundamental frequency of y with pYIN.
PitchTrack	類別	A pitch contour plus the voicing decision that makes it readable.
speech_ratio	函式	Fraction of the clip that vad calls speech, in [0, 1].
vad	函式	Find the spans of y that contain speech, in seconds.
vad_frames	函式	vad as one decision per STFT frame, True for speech.
vad_split	函式	The speech segments themselves, as waveforms.
vad_trim	函式	Trim to the first and last speech, keeping whatever is between them.

編輯與效果

完整說明在編輯與效果。

名稱	種類	用途（取自 docstring）
add_noise_snr	函式	Mix noise into y at the target signal-to-noise ratio (dB).
band_filter	函式	Butterworth filter of any of the four kinds.
bandpass	函式	Keep only low..high Hz.
bandstop	函式	Remove low..high Hz.
crossfade	函式	Join two clips through a short overlap, instead of butting them together.
DEFAULT_CROSSFADE	常數	Convert a string or number to a floating-point number, if possible.
fade	函式	Fade both ends in one call.
fade_in	函式	Ramp the first duration seconds up from silence.
fade_out	函式	Ramp the last duration seconds down to silence.
gain	函式	Apply a fixed gain in decibels.
highpass	函式	Remove everything below cutoff Hz — rumble, handling noise, DC drift.
lowpass	函式	Remove everything above cutoff Hz — hiss, and anything past the band a downstream model was trained on.
normalize	函式	Scale so the loudest sample sits at peak.
normalize_db	函式	Scale so the peak sits at dbfs decibels below full scale.
pitch_shift	函式	Shift pitch by n_steps semitones without changing duration.
random_gain	函式	Apply a random gain sampled uniformly from db_range dB.
remove_dc	函式	Subtract the mean, per channel.
reverb	函式	Cheap synthetic reverb via an exponentially-decaying comb filter: an echo every delay_ms, each decay times the one before.
reverse	函式	Play the waveform backwards.
spec_augment	函式	SpecAugment: mask random frequency bands and time spans of a spectrogram.
splice	函式	Join several clips end to end, crossfading every seam.
split_on_silence	函式	Split y into non-silent segments (each below-top_db gap is a cut).
time_stretch	函式	Speed up (rate>1) or slow down (rate<1) without changing pitch.
trim_silence	函式	Trim leading/trailing silence below top_db dB under peak.

房間與通道

完整說明在房間與通道。

名稱	種類	用途（取自 docstring）
a_law	函式	The same, through the A-law curve -- G.711 as Europe specifies it.
A_LAW_A	常數	Convert a string or number to a floating-point number, if possible.
apply_channel	函式	Run y through a whole link in the order a link does it.
apply_rir	函式	Put y in the room described by ir.
bit_depth	函式	Requantise y to bits, linearly.
CHANNEL_DEFAULTS	常數	dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object's (key, value) pairs dict(iterable) -> new dictionary initialized as if via: d = {} for k, v in iterable: d[k] = v dict(**kwargs) -> new dictionary initialized with the name=value pairs in the keyword argument list.
CHANNEL_PRESETS	常數	dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object's (key, value) pairs dict(iterable) -> new dictionary initialized as if via: d = {} for k, v in iterable: d[k] = v dict(**kwargs) -> new dictionary initialized with the name=value pairs in the keyword argument list.
channel_spec	函式	Normalise spec -- a preset name, a dict, or None -- to a dict.
channel_tag	函式	A short, stable description of spec, for a manifest or a filename.
dropouts	函式	Drop packets, the way a bad link does.
independent_burst	函式	The mean run length that independent loss at rate loss produces.
mu_law	函式	Push y through the G.711 mu-law companding curve and back.
MU_LAW_MU	常數	Convert a string or number to a floating-point number, if possible.
PACKET_MS	常數	Convert a string or number to a floating-point number, if possible.
Reverberation	類別	What a room does, measured from its impulse response.
rir	函式	A synthetic room impulse response: exponentially decaying noise.
rt60	函式	Measure a room from its impulse response, to ISO 3382-1.
schroeder_curve	函式	Backward-integrated energy decay of ir, in dB, starting at 0.
telephone	函式	Everything a phone line does, in the order it does it.
TELEPHONE_BAND	常數	Built-in immutable sequence.

資料集

完整說明在資料集。

名稱	種類	用途（取自 docstring）
Label	類別	A named span of a recording, in seconds.
load_labels	函式	Read labels written by save_labels, or by Audacity.
load_manifest	函式	Read a manifest written by save_manifest.
Pair	類別	One training example: a degraded file and the clean signal behind it.
save_labels	函式	Write labels as JSON, or as an Audacity label track.
save_manifest	函式	Write pairs to path as JSON.
split_pairs	函式	Shuffle once, then cut into (train, validation, test).
Synthesizer	類別	Mix clean audio with noise at the requested SNRs.

增強

完整說明在增強。

名稱	種類	用途（取自 docstring）
compare_enhancers	函式	Run several methods over the same clip and measure each one.
CUSTOM_NOISE	常數	str(object='') -> str str(bytes_or_buffer[, encoding[, errors]]) -> str
CustomEnhancer	類別	Anything that denoises, wrapped so it can be ranked with the rest.
DecisionDirected	類別	The a priori SNR estimator of Ephraim & Malah, as state plus an update.
enhance_folder	函式	Denoise every .wav file under src into dst.
EnhanceFolderResult	類別	What enhance_folder did.
EnhanceResult	類別	One run's output, with everything needed to rank it against another.
estimate_noise_psd	函式	Noise power spectrum for every frame of power (frames, bins).
make_gain_rule	函式	The per-frame gain function for one of RECURSIVE_METHODS.
make_noise_tracker	函式	A stateful tracker for method.
Method	類別	An enhancement method: what it is, and what it lets you turn.
METHODS	常數	dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object's (key, value) pairs dict(iterable) -> new dictionary initialized as if via: d = {} for k, v in iterable: d[k] = v dict(**kwargs) -> new dictionary initialized with the name=value pairs in the keyword argument list.
NOISE_ESTIMATORS	常數	dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object's (key, value) pairs dict(iterable) -> new dictionary initialized as if via: d = {} for k, v in iterable: d[k] = v dict(**kwargs) -> new dictionary initialized with the name=value pairs in the keyword argument list.
NoiseTracker	類別	A noise estimator as state plus one update, not a whole-file pass.
Param	類別	One tunable knob, described well enough to build a control from it.
RECURSIVE_METHODS	常數	Built-in immutable sequence.
spectral_enhance	函式	Enhance y with one of the methods in METHODS.
STREAMABLE_METHODS	常數	Built-in immutable sequence.
StreamEnhancer	類別	Denoise a stream block by block, with a one-frame delay.
trad_enhance	函式	Enhance waveform y using spectral subtraction with MCRA noise estimation and a Wiener-style gain function.
wavelet_low_pass_filter	函式	Soft-threshold the detail coefficients of a wavelet decomposition.

指標與對齊

完整說明在指標與對齊。

名稱	種類	用途（取自 docstring）
align	函式	Return ref and deg trimmed to the region they share, lined up.
Alignment	類別	Where two recordings sit relative to each other.
AudioEvaluate	類別	Evaluate enhanced audio against clean references.
check_metrics_install	函式	Report availability of the optional metric back-ends.
eval_metrics	函式	Score a single (reference, degraded) pair with every available metric.
find_delay	函式	Measure how far deg sits after ref, to the nearest sample.
MIN_CORRELATION	常數	Convert a string or number to a floating-point number, if possible.
pesq	函式	PESQ of deg against ref (ITU-T P.862 / P.862.2).
score	函式	Score deg against ref: {metric: value}.
sdi	函式	Speech distortion index: sum((ref - deg)**2) / sum(ref**2).
segmental_snr	函式	Frame-averaged (segmental) SNR in dB, clipped to [-10, 35] per frame.
si_sdr	函式	Scale-invariant SDR (dB) — immune to overall gain differences.
snr	函式	Signal-to-noise ratio (dB) of deg relative to ref (pure numpy).

即時音訊

完整說明在即時音訊。

名稱	種類	用途（取自 docstring）
AudioBuffer	類別	Bounded FIFO of audio frames shared between producer/consumer threads.
check_device	函式	List system devices and probe an input;
CheckDevice	類別	Query/validate PortAudio devices.
FetchBuffer	類別	Collect a fixed number of frames from an AudioBuffer in a background thread.
is_available	函式	Smoke-test the terminal colors and the audio input stack.
list_devices	函式	Devices as sounddevice sees them.
LocalStreamReader	類別	Stream from a local microphone into an AudioBuffer, with optional live monitoring through the default output device.
PA_FLOAT32	常數	int([x]) -> integer int(x, base=10) -> integer
PA_INT16	常數	int([x]) -> integer int(x, base=10) -> integer
play_audio	函式	Play audio from a file path, an opened wave.Wave_read, or an ndarray.
record	函式	Blocking record for seconds via sounddevice (modern backend).
Recorder	類別	Blocking recorder for a fixed duration.
RemoteStreamReader	類別	Receive an audio stream over TCP and feed it into an AudioBuffer (optionally playing it back locally).
StreamRecorder	類別	Record for as long as you like, and watch the level while it happens.
wave_decode	函式	Raw int16 bytes -> array of shape (chunk_length, channels).
wave_encode	函式	Multi-channel array -> interleaved int16 bytes.

設定與錯誤

完整說明在設定與錯誤。

名稱	種類	用途（取自 docstring）
AudioIOError	類別	Reading or writing an audio file failed.
DependencyError	類別	An optional dependency is required for this operation but not installed.
DeviceError	類別	An audio input/output device could not be opened or configured.
FeatureError	類別	A feature, effect, room, channel or enhancement step refused its input or its settings: the wrong shape of audio, a parameter out of range, a clip too short to frame.
KC	類別	Attribute-style config with YAML load/save and audio-centric defaults.
KudioConfig	類別	Attribute-style config with YAML load/save and audio-centric defaults.
KudioError	類別	Base class for every error raised by kudio.
SynthesisError	類別	Noisy-data synthesis failed.
UnsupportedFormatError	類別	The file type / dtype is not supported.

工具

完整說明在工具。

名稱	種類	用途（取自 docstring）
find_duplicate	函式	Return the Counter dict of file_list, logging duplicated entries.
lowercase	函式	The strings in lst, lower-cased.
map_waves	函式	Apply fn to every item, optionally across workers processes.
Timer	類別	Decorator: run func in a daemon thread, optionally joining for at most time_delay seconds.
timer	函式	Decorator: log the wall-clock run time of func.
uppercase	函式	The strings in lst, upper-cased.

舊名稱

這些名稱還能匯入，但每次呼叫都會發出警告，並指出新名稱。之後的版本會把它們移除。

舊名稱	新名稱
blu	kudio.util.blu
color_test	kudio.util.color_test
concat_logspec	logspec_from_files
concat_logspec_	logspec_from_files
concat_mfcc	mfcc_from_files
concat_mfcc_	mfcc_from_files
contextual_LogSpectrogram	logspec_from_files
cya	kudio.util.cya
dkblu	kudio.util.dkblu
dkcya	kudio.util.dkcya
dkgre	kudio.util.dkgre
dkpur	kudio.util.dkpur
dkred	kudio.util.dkred
dkyel	kudio.util.dkyel
error	kudio.util.error
f2s	file_to_spectrogram
fGre	kudio.util.fGre
fRed	kudio.util.fRed
gre	kudio.util.gre
h1	kudio.util.h1
info	kudio.util.info
pur	kudio.util.pur
red	kudio.util.red
s2w	save_spectrogram_as_wave
spec2wav	save_spectrogram_as_wave
spec2wavform	spectrogram_to_waveform
upercase	uppercase
w2mfcc	mfcc
w2s	waveform_to_spectrogram
warn	kudio.util.warn
wav2mel	melspectrogram
wav2mfcc	mfcc
wav2spec	file_to_spectrogram
wavform2spec	waveform_to_spectrogram
yel	kudio.util.yel