# Python API catalog

Search classes and functions by name or module. Documentation is generated from the Python source without importing training dependencies.

- [StreamingConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.StreamingConfig): Config for subject-level streaming dataset mode. — kind: class — category: configs
- [CacheConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.CacheConfig): Config for TFRecord cache dataset mode. — kind: class — category: configs
- [FilterConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.FilterConfig): Config for bandpass filtering of input or target signals. — kind: class — category: configs
- [AugmentationConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.AugmentationConfig): Configurable augmentation and abstention regimes applied in training. — kind: class — category: configs
- [SyntheticMixConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.SyntheticMixConfig): Config for synthetic ECG mixing during training. — kind: class — category: configs
- [DerivativeLossConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.DerivativeLossConfig): First-derivative (smoothness) penalty on reconstruction. — kind: class — category: configs
- [FilteredLossConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.FilteredLossConfig): Lowpass-filtered MSE to focus on frequencies the model can represent. — kind: class — category: configs
- [SpectralLossConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.SpectralLossConfig): Multi-scale spectral loss for frequency-domain reconstruction quality. — kind: class — category: configs
- [DwtLossConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.DwtLossConfig): Frequency-weighted MSE via Haar DWT subbands. — kind: class — category: configs
- [TransformConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.TransformConfig): Signal transform domain for compression input. — kind: class — category: configs
- [DataConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.DataConfig): Data loading and preprocessing configuration. — kind: class — category: configs
- [ModelConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.ModelConfig): RVQ autoencoder model architecture configuration. — kind: class — category: configs
- [LrScheduleConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.LrScheduleConfig): Learning rate schedule configuration. — kind: class — category: configs
- [BetaAnnealConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.BetaAnnealConfig): Schedule for the RVQ commitment-loss weight (beta). — kind: class — category: configs
- [RvqPrefixLossConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.RvqPrefixLossConfig): Auxiliary coarse-to-fine supervision for multi-level RVQ prefixes. — kind: class — category: configs
- [LabelTrustConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.LabelTrustConfig): Reference-free label-trust loss weighting (``n0``-aware strong/weak labels). — kind: class — category: configs
- [TrainingConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.TrainingConfig): Training hyperparameters and schedule configuration. — kind: class — category: configs
- [BandMetricsConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.BandMetricsConfig): Band-limited evaluation configuration. — kind: class — category: configs
- [StitchingEvalConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.StitchingEvalConfig): Long-recording stitching evaluation. — kind: class — category: configs
- [EvaluationConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.EvaluationConfig): Post-training evaluation configuration. — kind: class — category: configs
- [WandbConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.WandbConfig): Weights \& Biases logging configuration. — kind: class — category: configs
- [OutputConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.OutputConfig): Output directory and logging configuration. — kind: class — category: configs
- [EcgRvqConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ecg_rvq/#compressionkit.configs.ecg_rvq.EcgRvqConfig): Top-level configuration for ECG RVQ training pipeline. — kind: class — category: configs
- [StreamingConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.StreamingConfig): Config for subject-level streaming dataset mode. — kind: class — category: configs
- [CacheConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.CacheConfig): Config for TFRecord cache dataset mode. — kind: class — category: configs
- [SyntheticMixConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.SyntheticMixConfig): Config for synthetic PPG mixing during in-memory training. — kind: class — category: configs
- [FilterConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.FilterConfig): Config for bandpass filtering of input or target signals. — kind: class — category: configs
- [TransformConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.TransformConfig): Signal transform domain for compression input. — kind: class — category: configs
- [AugmentationConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.AugmentationConfig): Tier-1 PPG augmentation settings for cross-domain robustness. — kind: class — category: configs
- [UnifiedSourceConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.UnifiedSourceConfig): One source in a unified multi-source training mix. — kind: class — category: configs
- [UnifiedCacheConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.UnifiedCacheConfig): Config for unified per-source TFRecord cache dataset mode. — kind: class — category: configs
- [DataConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.DataConfig): Data loading and preprocessing configuration. — kind: class — category: configs
- [ModelConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.ModelConfig): RVQ autoencoder model architecture configuration. — kind: class — category: configs
- [LrScheduleConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.LrScheduleConfig): Learning rate schedule configuration. — kind: class — category: configs
- [DerivativeLossConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.DerivativeLossConfig): First-derivative (smoothness) penalty on reconstruction. — kind: class — category: configs
- [SpectralLossConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.SpectralLossConfig): Multi-scale spectral loss for frequency-domain reconstruction quality. — kind: class — category: configs
- [TrainingConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.TrainingConfig): Training hyperparameters and schedule configuration. — kind: class — category: configs
- [BandMetricsConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.BandMetricsConfig): Band-limited evaluation configuration. — kind: class — category: configs
- [PhysiokitMetricsConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.PhysiokitMetricsConfig): PhysioKit HR/HRV evaluation configuration. — kind: class — category: configs
- [LongRecordingEvalConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.LongRecordingEvalConfig): Long-recording overlap-add evaluation for clinically meaningful HR/HRV. — kind: class — category: configs
- [EvaluationConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.EvaluationConfig): Post-training evaluation configuration. — kind: class — category: configs
- [WandbConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.WandbConfig): Weights \& Biases logging configuration. — kind: class — category: configs
- [OutputConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.OutputConfig): Output directory and logging configuration. — kind: class — category: configs
- [PpgRvqConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/configs/ppg_rvq/#compressionkit.configs.ppg_rvq.PpgRvqConfig): Top-level configuration for PPG RVQ training pipeline. — kind: class — category: configs
- [load_ppg_signal](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.load_ppg_signal): Load a single PPG signal from an EDF file, resampling to *target_rate*. — kind: function — category: datasets
- [load_ppg_dataset](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.load_ppg_dataset): Load multiple EDF files into a stacked numpy array ``\[N, num_samples\]``. — kind: function — category: datasets
- [load_ppg_splits](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.load_ppg_splits): Load train/val/test numpy arrays of resampled PPG signals. — kind: function — category: datasets
- [load_ppg_file_splits](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.load_ppg_file_splits): Return train/val/test EDF file splits for subject-level separation. — kind: function — category: datasets
- [bandpass_filter_batch](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.bandpass_filter_batch): Apply a physiokit bandpass filter across a batch of signals. — kind: function — category: datasets
- [build_ppg_tfrecord_cache](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.build_ppg_tfrecord_cache): Build a subject-split TFRecord cache for PPG windows. — kind: function — category: datasets
- [make_ppg_inmemory_dataset](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.make_ppg_inmemory_dataset): Build a tf.data pipeline from in-memory numpy arrays. — kind: function — category: datasets
- [make_ppg_tfrecord_dataset](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.make_ppg_tfrecord_dataset): Build dataset from TFRecord windows compatible with RVQ trainer. — kind: function — category: datasets
- [make_ppg_stream_dataset](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.make_ppg_stream_dataset): Build a streaming tf.data dataset that samples random windows per subject. — kind: function — category: datasets
- [collect_random_samples](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/datasets/ppg/#compressionkit.datasets.ppg.collect_random_samples): Grab a subset of (input, target) tensors from a tf.data pipeline. — kind: function — category: datasets
- [save_sample_artifacts](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/evaluation/artifacts/#compressionkit.evaluation.artifacts.save_sample_artifacts): Save per-sample CSV, plot, and compute metrics for one evaluation sample. — kind: function — category: evaluation
- [PRD](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/evaluation/metrics/#compressionkit.evaluation.metrics.PRD): Percent RMS difference metric with optional energy normalization. — kind: class — category: evaluation
- [TruePRD](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/evaluation/metrics/#compressionkit.evaluation.metrics.TruePRD): Normalized PRD metric (convenience alias). — kind: class — category: evaluation
- [compute_signal_metrics](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/evaluation/metrics/#compressionkit.evaluation.metrics.compute_signal_metrics): Compute scalar reconstruction metrics on two aligned signals. — kind: function — category: evaluation
- [compute_ppg_physiokit_metrics](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/evaluation/metrics/#compressionkit.evaluation.metrics.compute_ppg_physiokit_metrics): Compute HR/HRV metrics for one PPG signal using physiokit. — kind: function — category: evaluation
- [summarize_ppg_peak_alignment](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/evaluation/metrics/#compressionkit.evaluation.metrics.summarize_ppg_peak_alignment): Compare PPG pulse peak timing between paired original/reconstructed signals. — kind: function — category: evaluation
- [summarize_physiokit_alignment](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/evaluation/metrics/#compressionkit.evaluation.metrics.summarize_physiokit_alignment): Compare physiokit HR/HRV metrics between original and reconstructed signals. — kind: function — category: evaluation
- [compute_ecg_hr_hrv](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/evaluation/metrics/#compressionkit.evaluation.metrics.compute_ecg_hr_hrv): Detect R-peaks and compute HR/HRV from a single ECG signal. — kind: function — category: evaluation
- [summarize_ecg_alignment](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/evaluation/metrics/#compressionkit.evaluation.metrics.summarize_ecg_alignment): Compare ECG HR/HRV/peak-timing between paired original and reconstructed signals. — kind: function — category: evaluation
- [export_encoder_tflite](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/export/tflite/#compressionkit.export.tflite.export_encoder_tflite): Export encoder to INT8 TFLite and C header using helia_edge. — kind: function — category: export
- [export_decoder_tflite](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/export/tflite/#compressionkit.export.tflite.export_decoder_tflite): Export decoder to INT8 TFLite and C header. — kind: function — category: export
- [export_denoiser_tflite](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/export/tflite/#compressionkit.export.tflite.export_denoiser_tflite): Export a wavelet-gain denoiser to INT8 TFLite and C header. — kind: function — category: export
- [build_decoder_2d](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/decoder/#compressionkit.models.decoder.build_decoder_2d): Build a configurable decoder that mirrors the encoder stages. — kind: function — category: models
- [build_decoder_2d_ssm](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/decoder/#compressionkit.models.decoder.build_decoder_2d_ssm): Decoder that mixes time with diagonal SSM blocks at every spatial scale. — kind: function — category: models
- [build_hierarchical_decoder_2d](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/decoder/#compressionkit.models.decoder.build_hierarchical_decoder_2d): Build a coarse + residual/detail decoder for per-level RVQ tensors. — kind: function — category: models
- [build_hierarchical_adaptor_decoder_2d](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/decoder/#compressionkit.models.decoder.build_hierarchical_adaptor_decoder_2d): Build a summed-latent decoder with per-level residual adaptors. — kind: function — category: models
- [build_decoder_2d_spatial](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/decoder/#compressionkit.models.decoder.build_decoder_2d_spatial): Build a 2-D spatial decoder that mirrors the spatial encoder. — kind: function — category: models
- [build_encoder_2d](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/encoder/#compressionkit.models.encoder.build_encoder_2d): Build a configurable encoder with ``2**num_stages`` downsampling. — kind: function — category: models
- [build_encoder_2d_invres](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/encoder/#compressionkit.models.encoder.build_encoder_2d_invres): Encoder using inverted-residual blocks with optional causal padding. — kind: function — category: models
- [build_encoder_2d_spatial](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/encoder/#compressionkit.models.encoder.build_encoder_2d_spatial): Build a 2-D spatial encoder for spectrogram inputs. — kind: function — category: models
- [PrefixSupervisedVQAutoencoder](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/rvq_autoencoder/#compressionkit.models.rvq_autoencoder.PrefixSupervisedVQAutoencoder): VQ autoencoder with auxiliary coarse-to-fine RVQ prefix losses. — kind: class — category: models
- [HierarchicalRVQAutoencoder](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/rvq_autoencoder/#compressionkit.models.rvq_autoencoder.HierarchicalRVQAutoencoder): RVQ autoencoder that routes per-level RVQ tensors to the decoder. — kind: class — category: models
- [build_rvq_autoencoder](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/rvq_autoencoder/#compressionkit.models.rvq_autoencoder.build_rvq_autoencoder): Build encoder, bottleneck, decoder, and composite VQAutoencoder. — kind: function — category: models
- [build_rvq_autoencoder_2d_spatial](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/rvq_autoencoder/#compressionkit.models.rvq_autoencoder.build_rvq_autoencoder_2d_spatial): Build 2-D spatial RVQ autoencoder for STFT spectrograms. — kind: function — category: models
- [compute_compression_stats](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/models/rvq_autoencoder/#compressionkit.models.rvq_autoencoder.compute_compression_stats): Compute compression ratio and related statistics. — kind: function — category: models
- [build_preprocessor](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/preprocessing/ppg/#compressionkit.preprocessing.ppg.build_preprocessor): Create preprocessing pipeline: random crop + layer normalization. — kind: function — category: preprocessing
- [build_augmenter](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/preprocessing/ppg/#compressionkit.preprocessing.ppg.build_augmenter): Create augmentation pipeline. — kind: function — category: preprocessing
- [generate_synthetic_ppg_batch](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/preprocessing/ppg/#compressionkit.preprocessing.ppg.generate_synthetic_ppg_batch): Generate synthetic PPG segments via physiokit. — kind: function — category: preprocessing
- [EncodedFrame](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/base/#compressionkit.runtime.base.EncodedFrame): `dataclass` — kind: class — category: runtime
- [Codec](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/base/#compressionkit.runtime.base.Codec): Uniform runtime interface for compressionKIT codecs. — kind: class — category: runtime
- [RVQCodec](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/codec/#compressionkit.runtime.codec.RVQCodec): Lightweight RVQ autoencoder codec using LiteRT for inference. — kind: class — category: runtime
- [HybridSpihtCodec](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/hybrid/#compressionkit.runtime.hybrid.HybridSpihtCodec): Learned-denoiser + SPIHT runtime loaded from a deploy directory. — kind: class — category: runtime
- [resolve_deploy_dir](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/loader/#compressionkit.runtime.loader.resolve_deploy_dir): Return a local deploy directory, downloading from HF if needed. — kind: function — category: runtime
- [load_codec](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/loader/#compressionkit.runtime.loader.load_codec): Hydrate a codec from a local deploy directory or HF repo id. — kind: function — category: runtime
- [EntropyPrior](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/prior/#compressionkit.runtime.prior.EntropyPrior): Lightweight entropy prior using a causal CNN TFLite model. — kind: class — category: runtime
- [SpihtCodecConfig](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/spiht/#compressionkit.runtime.spiht.SpihtCodecConfig): Typed codec parameters for SPIHT. — kind: class — category: runtime
- [SpihtCodec](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/spiht/#compressionkit.runtime.spiht.SpihtCodec): Thin runtime wrapper around :class:`SpihtAcCodec`. — kind: class — category: runtime
- [CompressionResult](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/two_stage/#compressionkit.runtime.two_stage.CompressionResult): `dataclass` — kind: class — category: runtime
- [TwoStageCodec](https://ambiqai.github.io/compressionkit/reference/api/compressionkit/runtime/two_stage/#compressionkit.runtime.two_stage.TwoStageCodec): Two-stage compression: RVQ codec + entropy-coded prior. — kind: class — category: runtime