Skip to content

Content is What Remains: Invariant Speech Tokenization from Parallel Utterances

Jul 2026 · arXiv.org · Vol abs/2607.19033 · 2 citations · 42 references
Computer Science

TL;DR

PINT (Parallel INvariant Tokenization), which fine-tunes an SSL encoder with alignment losses across parallel utterances and augmentations to distill this shared residual of linguistic content, and shows a 98.7% relative reduction in speaker probe accuracy.

Abstract

Discrete speech tokenizers aim to disentangle semantic from acoustic information, yet targets from self-supervised learning (SSL) models like HuBERT retain non-linguistic variation: speaker identity, prosody, and channel conditions leak into the tokens, inflating entropy. Our key insight is that when enough speakers utter the same words under varying conditions, linguistic content is the only shared factor. We propose PINT (Parallel INvariant Tokenization), which fine-tunes an SSL encoder with alignment losses across parallel utterances and augmentations to distill this shared residual. PINT collapses identical words onto consistent token sequences, drastically reducing conditional entropy. Unlike ASR text, PINT tokens preserve frame-level temporal grounding and serve as drop-in semantic targets for audio codecs. Experiments show a 98.7% relative reduction in speaker probe accuracy (93.1% to 1.2%), a 42% lower ABX error rate, and 27-30% lower LM perplexity versus baselines, confirming that the right invariance is key to efficient learning.

View source

Similar papers

Preprint Aug 2026

Speaker-Normalized Semantic Speech Tokens via Iterative S2U-T2U Refinement

Semantic speech tokens should preserve linguistic content while suppressing speaker- and duration-dependent variation inherited from acoustic inputs. We propose Iterative Semantic Token Purification (ISTP), an alternating speech-to-unit (S2U) and text-to-unit (T2U) training procedure guided by text predictability. Starting from an initial S2U tokenizer, each iteration trains a T2U model on its deduplicated token sequences. The decoded T2U predictions then serve as connectionist temporal classification targets for a newly initialized S2U model, whose outputs supervise the next T2U model. This cycle progressively aligns the two token generators and biases the token space toward information recoverable from text. Experiments on Mandarin and English show substantially improved S2U--T2U agreement. Independently trained de-tokenizers further show that the refined S2U and T2U tokens retain sufficient content for high-intelligibility voice conversion and text-to-speech synthesis. In voice conversion, the generated speaking rate follows the reference more closely. The refined tokens also exhibit substantially improved cross-speaker consistency and reduced probe-recoverable speaker information.

Hanlin Zhang, Da-Xin Tan, De-Hua Tao et al. · 0 citations
Preprint Aug 2026

Phoenix TTS: High-Fidelity Synthesis and Voice Conversion via Flow-Matching-Driven Speech Tokenization

This work proposes Phoenix TTS, a unified framework that tightly couples representation learning with generative acoustic modeling and is optimized to reconstruct self-supervised features to maintain semantic richness, while simultaneously receiving direct supervision from a Flow Matching training loss.

Peijie Chen, Zhuanling Zha, Zhipeng Nie et al. · 0 citations
#machine learning Preprint Sep 2026

TontaubeV1: Streaming Text-to-Speech with Hierarchical Codec Modeling and Bounded Context

Text-to-speech systems often face a trade-off between natural prosody and efficient inference: higher perceptual quality typically comes at increased computational cost and latency. We present TontaubeV1, a model that preserves natural prosody while enabling streaming from a single consumer GPU. Speech is encoded by the hierarchical DualCodec representation at 12.5 Hz, which separates a semantic stream from successive acoustic refinements. Our design assumes that prosodic structure is largely established when the semantic stream is generated, and allocates capacity accordingly: a Qwen3-1.7B-derived transformer predicts that stream and thereby the utterance duration, while three progressively smaller Qwen3-0.6B-derived transformers each add one acoustic refinement. Text is tokenized per character rather than by subword. Paired text and audio markers at shared positions support long-form generation with bounded context, and overlapping DualCodec reconstructions are mapped into the VibeVoice acoustic latent space and decoded causally, enabling streaming despite DualCodec's noncausal decoder. The model accepts up to one minute of reference audio for voice conditioning and is designed primarily for English and German, with additional multilingual support. The four predictors total 2.9B parameters; on a single RTX 5090 the streaming path reaches approximately 200 ms to first audio. In separate non-streaming measurements, the end-to-end real-time factor (RTF) is 0.08 for one input and the aggregate RTF is 0.02 across eight concurrent inputs. On our LLM-as-a-judge audiobook-reading benchmark, TontaubeV1 matches ElevenLabs Flash v2.5 and outperforms Fish Audio S2 Pro, the April 2026 Gradium API, and Cartesia Sonic 3 on prosody. The model weights are released on Hugging Face under the Tontaube Community Model License 1.0.

Fritz Cremer, Jonathan Cremer · 0 citations
Preprint Aug 2026

Confucius4-TTS: Transcript-Free Cross-Lingual Zero-Shot TTS with a Learnable Speaker Encoder

A multilingual zero-shot TTS system that supports 14 languages and performs both intra-lingual and cross-lingual reference cloning without requiring transcripts of audio prompts, which achieves the best average overall rank in human evaluation among recent open-source and commercial systems.

Hua-Xuan Wang, Hui-Min Wang, Ruiyu Zhang et al. · 0 citations
Preprint Aug 2026

Evaluating Pre-trained Speech Encoders for Spontaneous Speech Detection and Out of Domain Synthetic Speech Generalisation in Indic Languages

Centroid analysis shows that out-of-domain generalisation is predicted by a training system's proximity to unseen TTS embeddings, not its distance from natural speech, a finding with direct implications for training data selection in real-world deepfake detectors.

Varun Rai, J. PavanKumar, Sujith Pulikodan et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.