Skip to content

api-reference.md

Latest commit

 

History

History
73 lines (52 loc) · 2.8 KB

api-reference.md

File metadata and controls

73 lines (52 loc) · 2.8 KB
Raw

API 参考

仅列出稳定的公开接口；带 _ 前缀的均为内部实现。

配置

sonoloc.config.SonolocConfig

共享参数数据类。关键字段：sample_rate、n_fft、hop_length、n_mels、 fmin、fmax、label_rate、sound_speed、array。

  • frames_per_second -> float
  • label_hop -> int
  • to_dict() / from_dict(d)
  • save(path) / load(path)：YAML 序列化。

IO 与几何

  • sonoloc.io.geometry.MicArray(name, positions) — n_mics、pairs()、 from_spherical(name, az_deg, el_deg, radius)。
  • tetrahedral_array(radius=0.042) -> MicArray
  • get_array(name) -> MicArray
  • sph2cart(az, el, r) / cart2sph(xyz)
  • sonoloc.io.audio.load_audio(path, sample_rate=None) -> (signal, sr)
  • save_audio(path, signal, sample_rate) / resample_signal(signal, orig, target)

特征

  • sonoloc.features.stft.stft(signal, n_fft, hop_length, ...) -> ndarray
  • sonoloc.features.logmel.LogMelExtractor(sample_rate, n_fft, n_mels, ...)
  • sonoloc.features.gcc.gcc_phat(sig, ref, ...) / estimate_tdoa(sig, ref)
  • sonoloc.features.intensity.intensity_vectors(foa_spec) / intensity_doa(foa_spec)
  • sonoloc.features.pipeline.FeaturePipeline(config, array) — 可调用。

定位

  • sonoloc.localization.srp_phat(signal, array, config, grid=None) -> (az_deg, el_deg)
  • sonoloc.localization.music(signal, array, config, n_sources=1) -> (az_deg, el_deg)
  • sonoloc.localization.make_grid(az_step, el_step, el_min, el_max) -> SphereGrid
  • sonoloc.localization.steering_vector(positions, directions, freqs, c)

检测

  • sonoloc.detection.pooling.pool(frame_probs, method, axis)
  • sonoloc.detection.mil.clip_labels(frame_probs, method, threshold)
  • median_filter(frame_probs, size) / frames_to_events(frame_active, hop_seconds)

标签

  • sonoloc.labels.accdoa.encode_accdoa(activity, azimuth, elevation)
  • decode_accdoa(accdoa, threshold=0.5) -> dict
  • encode_multi_accdoa(...) / decode_multi_accdoa(...)
  • sonoloc.labels.events.EventVocabulary(classes=None)

指标

  • sonoloc.metrics.segment_detection_scores(reference, prediction, frames_per_segment=10)
  • localization_scores(ref_active, ref_az, ref_el, pred_active, pred_az, pred_el)
  • seld_score(error_rate, f_score, localization_error, localization_recall)
  • aggregate_seld(detection, localization) -> dict

数据

  • sonoloc.data.scene.Scene / SoundEvent
  • sonoloc.data.simulate.simulate_scene(scene, array, config, snr_db=None, seed=0)
  • random_scene(n_classes, duration, sample_rate, n_events, seed)
  • sonoloc.data.noise.add_noise_at_snr(signal, noise, snr_db)

模型（需要 [torch]）

  • sonoloc.models.crnn.CRNN(in_channels, n_mels, ...)
  • sonoloc.models.seld_model.SeldModel(in_channels, n_classes, ...)
  • sonoloc.models.accdoa.ACCDOAHead(in_dim, n_classes)