Quantizing an intermediate layer of a pretrained audio model with residual vector quantization, and finetuning the model with task and codebook losses, preserves ASR and audio classification accuracy at bitrates near 200-500 bps.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Task-Specific Audio Coding for Machines: Machine-Learned Latent Features Are Codes for That Machine
Quantizing an intermediate layer of a pretrained audio model with residual vector quantization, and finetuning the model with task and codebook losses, preserves ASR and audio classification accuracy at bitrates near 200-500 bps.