compressionkit-ecg-4x-v1.0

A ECG signal compression codec using Residual Vector Quantization (RVQ), optimized for edge and wearable devices.

Model Details

• Modality: ECG
• Sample Rate: 256 Hz
• Compression Ratio: 4x
• Quantization: INT8
• RVQ Levels: 4
• Codebook Size: 256 entries × 16D
• Encoder Input: [None, 1, 512, 1]
• Encoder Output: [None, 1, 128, 16]

Quality Metrics

Fidelity & Robustness

Both fidelity yardsticks are reported so the codec is judged fairly: faithfulness is PRD vs the recorded (still-noisy) input, while truth fidelity is PRD vs clean ground truth. Lower is better.

Metric	Value
Truth PRD vs clean (%)	2.33
Truth PRD at native noise (%)	38.03
Faithful PRD vs input (%)	3.63
PRD degradation slope (PRD%/dB)	3.97
PRD at 0 dB SNR (%)	51.15
PRD at -6 dB SNR (%)	79.05
Pure-noise imprint autocorr	0.2496

Time Domain

PRD here is faithfulness (vs the recorded input); see Fidelity & Robustness above for the clean-truth and noise-regime view.

Metric	Mean	Median	P90
PRD vs input — faithfulness (%)	3.6256	3.2914	5.2169
RMSE	0.0349	0.0316	0.0495
Cosine Similarity	0.9992	0.9995	0.9997

Spectral

• Band Total Relative Error (median): 0.0427

Bitrate

• Codec CR (uniform): 4.0x
• Codec CR (learned prior): 9.73x

Encoder Precision Parity

Difference from FP32 reconstruction on a disjoint real-data holdout; lower is better.

Encoder	P90 PRD	Worst PRD	Status
FP16	0.38%	1.52%	recommended
INT16X8	1.53%	9.95%	recommended
INT8	4.16%	10.53%	recommended

Usage

Python (compressionkit runtime)

from compressionkit.runtime import RVQCodec

codec = RVQCodec.from_pretrained("Ambiq/compressionkit-ecg-4x-v1.0")

# Encode: float32 signal → RVQ indices
indices = codec.encode(signal)

# Decode: RVQ indices → reconstructed signal
recon = codec.decode(indices)

Local deployment directory

codec = RVQCodec("path/to/deploy/")

Files

File	Description
encoder_int8.tflite	INT8 quantized encoder (on-device)
encoder_float32.tflite	Float32 encoder for browser/server runtimes
encoder_fp16.tflite	FP16 encoder variant for supported edge runtimes
encoder_int16x8.tflite	INT16x8 encoder variant for supported edge runtimes
encoder.h	C header for encoder
encoder.keras	Float32 Python reference encoder (training/inspection use)
decoder_float32.tflite	Float32 decoder (server-side evaluation)
decoder_int8.tflite	INT8 decoder (optional, on-device)
decoder.keras	Float32 Python reference decoder (training/inspection use)
codebook.npz	RVQ codebook tables
codebook.h	C header for codebook
config.json	Deployment manifest
sample_stimulus.npz	Synthetic test data
demo_recordings.npz	10 real, quality-gated browser-demo recordings
demo_recordings_manifest.json	Recording provenance, license, and quality metadata
quality_scorecard.json	Full evaluation metrics

Dataset & License

Training data: PTB-XL (CC BY 4.0). Sample data may include excerpts under the original license terms. Demo recordings: MIT-BIH Arrhythmia Database v1.0.0 (source); real, quality-gated excerpts are released under ODC-By-1.0. https://opendatacommons.org/licenses/by/1-0/

Model weights are released under the Ambiq Model Weights License — deployment is restricted to Ambiq silicon devices. See LICENSE-MODEL-WEIGHTS.md for full terms.

Citation

@software{compressionkit,
  author = {Ambiq AI},
  title = {compressionKIT: Signal Compression for Edge AI},
  url = {https://github.com/AmbiqAI/compressionkit}
}