@inproceedings{gupta-etal-2026-audible,
title = "Am {I} Audible Now? Dissecting Prosody in Neural Audio Codecs",
author = "Gupta, Nikunj and
Rai, Shreshth and
Sharma, Kartik and
Vishwakarma, Dinesh K",
editor = "T.Y.S.S., Santosh and
Ghosh, Sreyan and
Duraiswami, Ramani and
Watanabe, Shinji and
Arora, Siddhant and
Kumar, Sonal and
Adi, Yossi and
Manocha, Dinesh and
Garcia Perera, Leibny Paola",
booktitle = "The Proceedings of Speech and Audio Language Models Workshop (2nd Edition)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.salma-1.9/",
pages = "121--133",
ISBN = "979-8-89176-496-5",
abstract = "The functional organization of prosody within the Residual Vector Quantization (RVQ) codebooks of neural audio codecs remains an open question. Prior work relies on correlational probing, showing that pitch features can be decoded from frozen codebook embeddings. Decodability, however, does not imply causal necessity: a codebook may encode pitch information the decoder ignores, while another may be essential despite containing little probe-recoverable signal. We address this gap through causal output-side ablation: for each codebook, we zero or mean-replace its discrete tokens, re-decode the waveform, and measure degradation in fundamental frequency $F_0$ (CREPE), energy (frame-wise RMS), and intelligibility (STOI) against the unablated reconstruction. Experiments span 200 LibriSpeech test-clean utterances under clean, 15, 5, and 0 dB SNR with MUSAN noise across multiple neural audio codecs. Across different architectures, we find a consistent two-tier causal structure: a codec-specific subset of codebooks is reconstruction-critical, while the remaining intelligibility-preserving codebooks exhibit a stable pitch-contribution hierarchy that persists under additive noise. These findings provide new evidence for how neural audio codecs causally organize prosodic information."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gupta-etal-2026-audible">
<titleInfo>
<title>Am I Audible Now? Dissecting Prosody in Neural Audio Codecs</title>
</titleInfo>
<name type="personal">
<namePart type="given">Nikunj</namePart>
<namePart type="family">Gupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shreshth</namePart>
<namePart type="family">Rai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kartik</namePart>
<namePart type="family">Sharma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dinesh</namePart>
<namePart type="given">K</namePart>
<namePart type="family">Vishwakarma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>The Proceedings of Speech and Audio Language Models Workshop (2nd Edition)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Santosh</namePart>
<namePart type="family">T.Y.S.S.</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sreyan</namePart>
<namePart type="family">Ghosh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ramani</namePart>
<namePart type="family">Duraiswami</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shinji</namePart>
<namePart type="family">Watanabe</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siddhant</namePart>
<namePart type="family">Arora</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sonal</namePart>
<namePart type="family">Kumar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yossi</namePart>
<namePart type="family">Adi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dinesh</namePart>
<namePart type="family">Manocha</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Leibny</namePart>
<namePart type="given">Paola</namePart>
<namePart type="family">Garcia Perera</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-496-5</identifier>
</relatedItem>
<abstract>The functional organization of prosody within the Residual Vector Quantization (RVQ) codebooks of neural audio codecs remains an open question. Prior work relies on correlational probing, showing that pitch features can be decoded from frozen codebook embeddings. Decodability, however, does not imply causal necessity: a codebook may encode pitch information the decoder ignores, while another may be essential despite containing little probe-recoverable signal. We address this gap through causal output-side ablation: for each codebook, we zero or mean-replace its discrete tokens, re-decode the waveform, and measure degradation in fundamental frequency F₀ (CREPE), energy (frame-wise RMS), and intelligibility (STOI) against the unablated reconstruction. Experiments span 200 LibriSpeech test-clean utterances under clean, 15, 5, and 0 dB SNR with MUSAN noise across multiple neural audio codecs. Across different architectures, we find a consistent two-tier causal structure: a codec-specific subset of codebooks is reconstruction-critical, while the remaining intelligibility-preserving codebooks exhibit a stable pitch-contribution hierarchy that persists under additive noise. These findings provide new evidence for how neural audio codecs causally organize prosodic information.</abstract>
<identifier type="citekey">gupta-etal-2026-audible</identifier>
<location>
<url>https://aclanthology.org/2026.salma-1.9/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>121</start>
<end>133</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Am I Audible Now? Dissecting Prosody in Neural Audio Codecs
%A Gupta, Nikunj
%A Rai, Shreshth
%A Sharma, Kartik
%A Vishwakarma, Dinesh K.
%Y T.Y.S.S., Santosh
%Y Ghosh, Sreyan
%Y Duraiswami, Ramani
%Y Watanabe, Shinji
%Y Arora, Siddhant
%Y Kumar, Sonal
%Y Adi, Yossi
%Y Manocha, Dinesh
%Y Garcia Perera, Leibny Paola
%S The Proceedings of Speech and Audio Language Models Workshop (2nd Edition)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-496-5
%F gupta-etal-2026-audible
%X The functional organization of prosody within the Residual Vector Quantization (RVQ) codebooks of neural audio codecs remains an open question. Prior work relies on correlational probing, showing that pitch features can be decoded from frozen codebook embeddings. Decodability, however, does not imply causal necessity: a codebook may encode pitch information the decoder ignores, while another may be essential despite containing little probe-recoverable signal. We address this gap through causal output-side ablation: for each codebook, we zero or mean-replace its discrete tokens, re-decode the waveform, and measure degradation in fundamental frequency F₀ (CREPE), energy (frame-wise RMS), and intelligibility (STOI) against the unablated reconstruction. Experiments span 200 LibriSpeech test-clean utterances under clean, 15, 5, and 0 dB SNR with MUSAN noise across multiple neural audio codecs. Across different architectures, we find a consistent two-tier causal structure: a codec-specific subset of codebooks is reconstruction-critical, while the remaining intelligibility-preserving codebooks exhibit a stable pitch-contribution hierarchy that persists under additive noise. These findings provide new evidence for how neural audio codecs causally organize prosodic information.
%U https://aclanthology.org/2026.salma-1.9/
%P 121-133
Markdown (Informal)
[Am I Audible Now? Dissecting Prosody in Neural Audio Codecs](https://aclanthology.org/2026.salma-1.9/) (Gupta et al., SALMA 2026)
ACL