SpeechCombine produces instruction-following SLMs via speech pre-training followed by direct weight combination with the text LLM instruction delta, without any speech instruction tuning.
Title resolution pending
11 Pith papers cite this work. Polarity classification is still indexing.
11
Pith papers citing it
citation-role summary
dataset 1
citation-polarity summary
roles
dataset 1polarities
use dataset 1representative citing papers
Presents the LibriQuote dataset of expressive audiobook speech and shows that fine-tuning or training TTS models on it improves expressivity and intelligibility.
UniVocal presents a text-context-only framework for speech-singing code-switching synthesis via two-stage curriculum learning and a synthetic data pipeline, claiming SOTA on a new benchmark.
Block-diffusion decoder with prior-calibrated scoring and early stopping produces streaming zero-shot TTS at quality comparable to AR and NAR baselines with lower real-time factor.
JASTIN is an instruction-driven audio evaluation system that achieves state-of-the-art correlation with human ratings on speech, sound, music, and out-of-domain tasks without task-specific retraining.
StableToken introduces a multi-branch architecture with bit-wise voting to create noise-robust semantic speech tokens, achieving lower Unit Edit Distance and better SpeechLLM robustness than prior single-path tokenizers.
Speech language models fail at reasoning about sentence stress but improve after fine-tuning on a new 17k-example synthetic dataset that varies stress to alter meaning.
CosyVoice 3 achieves better content consistency, speaker similarity, and prosody naturalness in zero-shot multilingual speech synthesis by scaling data to one million hours, model size to 1.5 billion parameters, and introducing a supervised multi-task speech tokenizer plus a differentiable reward模型.
A tag-based annotation scheme for non-verbal vocalizations in TTS data yields higher expressiveness (eMOS 4.20) and emotion recognition accuracy (78.8%) with minor naturalness trade-offs.
Raon-OpenTTS provides an open 510K-hour curated speech dataset and DiT-based TTS models up to 1B parameters that achieve competitive WER and speaker similarity on benchmarks versus closed models trained on millions of hours.
Voxtral TTS produces expressive multilingual speech from 3-second reference audio with a hybrid autoregressive-plus-flow-matching architecture and a new VQ-FSQ tokenizer, achieving 68.4% win rate over ElevenLabs in human evaluations.
citing papers explorer
-
Unlocking Speech-Text Compositional Powers: Instruction-Following Speech Language Models without Instruction Tuning
SpeechCombine produces instruction-following SLMs via speech pre-training followed by direct weight combination with the text LLM instruction delta, without any speech instruction tuning.
-
Computational Narrative Understanding for Expressive Text-to-Speech
Presents the LibriQuote dataset of expressive audiobook speech and shows that fine-tuning or training TTS models on it improves expressivity and intelligibility.
-
UniVocal: Unified Speech-Singing Code-Switching Synthesis
UniVocal presents a text-context-only framework for speech-singing code-switching synthesis via two-stage curriculum learning and a synthetic data pipeline, claiming SOTA on a new benchmark.
-
Chatterbox-Flash: Prior-Calibrated Block Diffusion for Streaming Zero-Shot TTS
Block-diffusion decoder with prior-calibrated scoring and early stopping produces streaming zero-shot TTS at quality comparable to AR and NAR baselines with lower real-time factor.
-
JASTIN: Aligning LLMs for Zero-Shot Audio and Speech Evaluation via Natural Language Instructions
JASTIN is an instruction-driven audio evaluation system that achieves state-of-the-art correlation with human ratings on speech, sound, music, and out-of-domain tasks without task-specific retraining.
-
StableToken: A Noise-Robust Semantic Speech Tokenizer for Resilient SpeechLLMs
StableToken introduces a multi-branch architecture with bit-wise voting to create noise-robust semantic speech tokens, achieving lower Unit Edit Distance and better SpeechLLM robustness than prior single-path tokenizers.
-
StressTest: Can YOUR Speech LM Handle the Stress?
Speech language models fail at reasoning about sentence stress but improve after fine-tuning on a new 17k-example synthetic dataset that varies stress to alter meaning.
-
CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training
CosyVoice 3 achieves better content consistency, speaker similarity, and prosody naturalness in zero-shot multilingual speech synthesis by scaling data to one million hours, model size to 1.5 billion parameters, and introducing a supervised multi-task speech tokenizer plus a differentiable reward模型.
-
Toward Natural Emotional Text-To-Speech System with Fine-Grained Non-Verbal Expression Control
A tag-based annotation scheme for non-verbal vocalizations in TTS data yields higher expressiveness (eMOS 4.20) and emotion recognition accuracy (78.8%) with minor naturalness trade-offs.
-
Raon-OpenTTS: Open Models and Data for Robust Text-to-Speech
Raon-OpenTTS provides an open 510K-hour curated speech dataset and DiT-based TTS models up to 1B parameters that achieve competitive WER and speaker similarity on benchmarks versus closed models trained on millions of hours.
-
Voxtral TTS
Voxtral TTS produces expressive multilingual speech from 3-second reference audio with a hybrid autoregressive-plus-flow-matching architecture and a new VQ-FSQ tokenizer, achieving 68.4% win rate over ElevenLabs in human evaluations.