mlboydaisuke / DAC-16kHz-LiteRT

huggingface.co
Total runs: 44
24-hour runs: 3
7-day runs: 6
30-day runs: 35
Model's Last Updated: August 15 2026
audio-to-audio

Introduction of DAC-16kHz-LiteRT

Model Details of DAC-16kHz-LiteRT

DAC (Descript Audio Codec) 16 kHz — LiteRT (CompiledModel GPU)

Descript Audio Codec running on-device on the LiteRT CompiledModel GPU (ML Drift). The convolutional encoder/decoder run on the GPU; the RVQ runs on CPU. 43:1 compression (1 s → 12×50 codes), RTF ≈ 0.82 (faster than real-time) on Pixel 8a.

Files
  • dac_16khz_encoder_fp16.tflite (43 MB) — audio[1,1,16000] → latent[1,1024,50] , GPU.
  • dac_16khz_deconly_zs_fp16.tflite (105 MB) — latent[1,1024,50] → audio , GPU.
  • dac_rvq.bin (1.2 MB) — RVQ weights (12 codebooks) for the CPU quantizer (float32 LE).
Pipeline
audio -> encoder.tflite (GPU) -> z -> RVQ.encode (CPU) -> codes[12,50]
      -> RVQ.decode (CPU) -> z_q -> decoder.tflite (GPU) -> audio
On-device (Pixel 8a, Tensor G3 — verified)

encoder 367/367 + decoder 398/398 nodes on the LiteRT GPU delegate ( LITERT_CL , 1 partition, no CPU fallback); warm RTF ~0.82; reconstruction corr 1.0 vs PyTorch DAC.

Why the split

The decoder's ConvTranspose1d are rewritten to a GPU-clean zero-stuff form (the real DAC's odd stride-5 transposed conv fails converter legalization, and TRANSPOSE_CONV is rejected by Mali). The RVQ uses EMBEDDING_LOOKUP + int64 indices (Mali-rejected) so it runs on CPU. So the float conv graph stays fully on the GPU.

Android sample + conversion/validation scripts: https://github.com/john-rocky/LiteRT-Models/tree/main/dac

License: MIT (Descript DAC).

Runs of mlboydaisuke DAC-16kHz-LiteRT on huggingface.co

44
Total runs
3
24-hour runs
9
3-day runs
6
7-day runs
35
30-day runs

More Information About DAC-16kHz-LiteRT huggingface.co Model

More DAC-16kHz-LiteRT license Visit here:

https://choosealicense.com/licenses/mit

DAC-16kHz-LiteRT huggingface.co

DAC-16kHz-LiteRT huggingface.co is an AI model on huggingface.co that provides DAC-16kHz-LiteRT's model effect (), which can be used instantly with this mlboydaisuke DAC-16kHz-LiteRT model. huggingface.co supports a free trial of the DAC-16kHz-LiteRT model, and also provides paid use of the DAC-16kHz-LiteRT. Support call DAC-16kHz-LiteRT model through api, including Node.js, Python, http.

mlboydaisuke DAC-16kHz-LiteRT online free

DAC-16kHz-LiteRT huggingface.co is an online trial and call api platform, which integrates DAC-16kHz-LiteRT's modeling effects, including api services, and provides a free online trial of DAC-16kHz-LiteRT, you can try DAC-16kHz-LiteRT online for free by clicking the link below.

mlboydaisuke DAC-16kHz-LiteRT online free url in huggingface.co:

https://huggingface.co/mlboydaisuke/DAC-16kHz-LiteRT

DAC-16kHz-LiteRT install

DAC-16kHz-LiteRT is an open source model from GitHub that offers a free installation service, and any user can find DAC-16kHz-LiteRT on GitHub to install. At the same time, huggingface.co provides the effect of DAC-16kHz-LiteRT install, users can directly use DAC-16kHz-LiteRT installed effect in huggingface.co for debugging and trial. It also supports api for free installation.

DAC-16kHz-LiteRT install url in huggingface.co:

https://huggingface.co/mlboydaisuke/DAC-16kHz-LiteRT

Url of DAC-16kHz-LiteRT

DAC-16kHz-LiteRT huggingface.co Url

Provider of DAC-16kHz-LiteRT huggingface.co

mlboydaisuke
ORGANIZATIONS

Other API from mlboydaisuke