Pith. sign in

REVIEW 40 cited by

Llasa: Scaling Train-Time and Inference-Time Compute for Llama-based Speech Synthesis

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2502.04128 v2 pith:IZURCTWC submitted 2025-02-06 eess.AS cs.AIcs.CLcs.MMcs.SD

Llasa: Scaling Train-Time and Inference-Time Compute for Llama-based Speech Synthesis

classification eess.AS cs.AIcs.CLcs.MMcs.SD
keywords computescalinginference-timespeechmodelmodelsllasallms
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Recent advances in text-based large language models (LLMs), particularly in the GPT series and the o1 model, have demonstrated the effectiveness of scaling both training-time and inference-time compute. However, current state-of-the-art TTS systems leveraging LLMs are often multi-stage, requiring separate models (e.g., diffusion models after LLM), complicating the decision of whether to scale a particular model during training or testing. This work makes the following contributions: First, we explore the scaling of train-time and inference-time compute for speech synthesis. Second, we propose a simple framework Llasa for speech synthesis that employs a single-layer vector quantizer (VQ) codec and a single Transformer architecture to fully align with standard LLMs such as Llama. Our experiments reveal that scaling train-time compute for Llasa consistently improves the naturalness of synthesized speech and enables the generation of more complex and accurate prosody patterns. Furthermore, from the perspective of scaling inference-time compute, we employ speech understanding models as verifiers during the search, finding that scaling inference-time compute shifts the sampling modes toward the preferences of specific verifiers, thereby improving emotional expressiveness, timbre consistency, and content accuracy. In addition, we released the checkpoint and training code for our TTS model (1B, 3B, 8B) and codec model publicly available.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 40 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. WavTTS: Towards High-Quality Zero-Shot TTS via Direct Raw Waveform Modeling

    eess.AS 2026-06 unverdicted novelty 8.0

    WavTTS is the first raw-waveform diffusion TTS model using DiT flow matching and multi-scale mel supervision that approaches SOTA latent zero-shot performance while beating prior end-to-end models.

  2. FlexiSLM: A Dynamic and Controllable Frame Rate Spoken Language Model

    cs.SD 2026-06 unverdicted novelty 7.0

    FlexiSLM is the first spoken language model supporting dynamic and controllable frame rates on speech input and output, outperforming fixed-rate 7B models at high quality and enabling faster inference at lower rates l...

  3. DTM-Codec: Dynamic Token Masking for VFR Speech Coding with Efficient Boundary Selection

    eess.AS 2026-06 unverdicted novelty 7.0

    DTM-Codec achieves better reconstruction quality and intelligibility than fixed-frame-rate neural speech codecs at matched total bitrate via dynamic token masking and Path Length Equalization for variable frame rates.

  4. One-Step Token-to-Waveform Generation with MeanFlow in Latent Space

    eess.AS 2026-06 unverdicted novelty 7.0

    MeanFlow applied in latent space enables true one-step Token2Wav generation with up to 17x RTF improvement and negligible quality loss versus multi-step baselines.

  5. Optimising Neural Speech Codecs for 300bps Communication using Reinforcement Learning

    cs.SD 2026-05 unverdicted novelty 7.0

    ClariCodec achieves 3.55% WER on LibriSpeech test-clean at 300 bps by RL fine-tuning the encoder for intelligibility, yielding a 23% relative WER reduction while preserving perceptual quality.

  6. AffectCodec: Emotion-Preserving Neural Speech Codec for Expressive Speech Modeling

    cs.SD 2026-05 unverdicted novelty 7.0

    AffectCodec is an emotion-guided neural speech codec that preserves emotional cues during quantization while maintaining semantic fidelity and prosodic naturalness.

  7. MINT-Bench: A Comprehensive Multilingual Benchmark for Instruction-Following Text-to-Speech

    eess.AS 2026-04 unverdicted novelty 7.0

    MINT-Bench is a new benchmark using hierarchical taxonomy, multi-stage data pipeline, and hybrid evaluation to assess instruction-following TTS systems, revealing major gaps in compositional and paralinguistic controls.

  8. Hierarchical Decoding for Discrete Speech Synthesis with Multi-Resolution Spoof Detection

    cs.SD 2026-03 unverdicted novelty 7.0

    MSpoof-TTS improves zero-shot discrete speech synthesis by integrating multi-resolution token-based spoof detection into a hierarchical decoding process that prunes low-quality candidates.

  9. JUST-DUB-IT: Video Dubbing via Joint Audio-Visual Diffusion

    cs.GR 2026-01 unverdicted novelty 7.0

    JUST-DUB-IT adapts a joint audio-visual diffusion model via LoRA to generate high-quality dubbed videos with translated audio and lip-synced facial motion.

  10. ReGen: Hierarchical Multi-Prompt Representation Generation for Efficient Waveform Diffusion Models

    cs.SD 2026-07 conditional novelty 6.5

    Hierarchical multi-prompt representation generation plus generalized flow matching yields high-quality single-stage waveform diffusion from 12.5 Hz latents and efficient LDM TTS.

  11. Unified Audio Intelligence Without Regressing on Text Intelligence

    cs.CL 2026-07 conditional novelty 6.0

    Audex unifies audio understanding and generation on a strong text MoE backbone with multi-stage SFT plus text-only Cascade RL, matching open SOTA audio scores while mostly retaining text capability.

  12. GRAFT: Grafted Reference Audio for Fine-grained Pronunciation in Zero-shot Text-to-Speech

    cs.LG 2026-07 accept novelty 6.0

    GRAFT splices a short spoken word sample into a neural codec TTS prompt and uses voice-conversion training so the model copies that pronunciation into any target voice, cutting target-word phoneme error 22-39%.

  13. FlowTTS-GRPO: Online Reinforcement Learning with Multi-Objective Reward Optimization for Flow-Matching Based Text-to-Speech

    eess.AS 2026-06 conditional novelty 6.0

    FlowTTS-GRPO fine-tunes open-source flow-matching TTS models with multi-objective online RL via ODE-to-SDE conversion, improving speaker similarity and quality on CosyVoice 3.0 and F5-TTS.

  14. ProsoCodec: Prosody-Oriented Speech Codec for Voice Conversion

    eess.AS 2026-06 unverdicted novelty 6.0

    ProsoCodec models prosody as a conditional residual in a speech codec via text and speaker prefix conditioning, yielding improved prosody preservation and less timbre leakage in voice conversion experiments.

  15. Reliable Neural-Codec Text-to-Speech by ASR Self-Verification and Distillation: Near-Zero Catastrophic Failures Across Models and Codecs

    cs.SD 2026-06 unverdicted novelty 6.0

    ASR self-verification via best-of-N sampling eliminates observed catastrophic failures in multiple neural-codec TTS models, with distillation transferring most of the robustness to single-shot decoding.

  16. Self-Guidance: Enhancing Neural Codecs via Decoder Manifold Alignment

    cs.SD 2026-06 unverdicted novelty 6.0

    Self-guidance adds a lightweight feature-mapping loss to align decoder manifolds in VQ-VAE speech codecs, raising reconstruction metrics and allowing 4x codebook reduction with no fidelity loss.

  17. TLDR: Compressing Audio Tokens for Efficient Autoregressive Text-to-Speech

    cs.SD 2026-06 unverdicted novelty 6.0

    TLDR groups codec tokens into patches for patch-level autoregressive modeling in pretrained TTS systems, yielding 1.8x speedup and 75% KV-cache reduction at patch size 4.

  18. dots.tts Technical Report

    cs.SD 2026-06 unverdicted novelty 6.0

    dots.tts reports SOTA benchmark results on Seed-TTS-Eval and other tests via continuous latent-space autoregressive modeling with three listed innovations and code release.

  19. CleanCodec: Efficient and Robust Speech Tokenization via Perceptually Guided Encoding

    cs.SD 2026-06 unverdicted novelty 6.0

    CleanCodec reframes audio tokenization as a selective information bottleneck to encode only perceptually important features at 12.5 tokens per second, outperforming prior codecs in efficiency, speaker similarity, and ...

  20. SemaVoice: Semantic-Aware Continuous Autoregressive Speech Synthesis

    eess.AS 2026-05 unverdicted novelty 6.0

    SemaVoice adds SFM-guided alignment to refine continuous speech representations in autoregressive TTS, reporting 1.71% English WER on Seed-TTS and competitiveness with open-source SOTA.

  21. Break-the-Beat! Controllable MIDI-to-Drum Audio Synthesis

    cs.SD 2026-05 unverdicted novelty 6.0

    Break-the-Beat! renders drum MIDI audio that matches the timbre of a reference clip by fine-tuning a text-to-audio model with a content encoder and hybrid conditioning on a new paired dataset.

  22. Reducing Linguistic Hallucination in LM-Based Speech Enhancement via Noise-Invariant Acoustic-Semantic Distillation

    eess.AS 2026-05 unverdicted novelty 6.0

    L3-SE reduces linguistic hallucination in LM-based speech enhancement by distilling noise-invariant acoustic-semantic representations from noisy inputs to condition an autoregressive decoder-only language model.

  23. RTCFake: Speech Deepfake Detection in Real-Time Communication

    cs.SD 2026-04 unverdicted novelty 6.0

    RTCFake is the first large-scale dataset of real-time communication speech deepfakes paired with offline versions, paired with a phoneme-guided consistency learning method that improves cross-platform and noise-robust...

  24. OmniVoice: Towards Omnilingual Zero-Shot Text-to-Speech with Diffusion Language Models

    cs.CL 2026-04 unverdicted novelty 6.0

    OmniVoice introduces a diffusion language model-style non-autoregressive TTS system that directly maps text to multi-codebook acoustic tokens, scaling zero-shot synthesis to over 600 languages with SOTA results on mul...

  25. Qwen3-TTS Technical Report

    cs.SD 2026-01 unverdicted novelty 6.0

    Qwen3-TTS delivers state-of-the-art multilingual TTS performance with 3-second voice cloning, description control, and ultra-low-latency streaming via dual tokenizers and a dual-track LM architecture trained on over 5...

  26. Two-Dimensional Quantization for Geometry-Aware Audio Coding

    cs.SD 2025-12 unverdicted novelty 6.0

    Q2D2 uses 2D geometric grid projections to quantize feature pairs in neural audio codecs, yielding implicit codebooks that improve efficiency and utilization over RVQ, VQ, and FSQ while maintaining reconstruction quality.

  27. Unified Audio Intelligence Without Regressing on Text Intelligence

    cs.CL 2026-07 conditional novelty 5.0

    A unified 30B MoE audio-text LLM achieves state-of-the-art audio understanding, generation, and speech tasks while preserving text reasoning comparable to its text-only backbone.

  28. FlowTTS-GRPO: Online Reinforcement Learning with Multi-Objective Reward Optimization for Flow-Matching Based Text-to-Speech

    eess.AS 2026-06 unverdicted novelty 5.0

    FlowTTS-GRPO applies online RL with weighted multi-objective rewards to flow-matching TTS models via ODE-to-SDE conversion, reporting gains in speaker similarity and perceptual quality on CosyVoice 3.0 and F5-TTS.

  29. FlashTTS: Fast Streaming TTS with MTP Acceleration and X-pred Mean Flow Distillation

    eess.AS 2026-06 unverdicted novelty 5.0

    FlashTTS delivers a streaming TTS system using multi-track input processing and X-pred mean flow matching to reach 325 ms latency in two function evaluations while retaining zero-shot voice cloning.

  30. VoxCPM2 Technical Report

    cs.SD 2026-06 unverdicted novelty 5.0

    VoxCPM2 scales hierarchical continuous-latent speech modeling to 2B parameters and over 2M hours of multilingual data, unifying voice cloning, style control, and continuation in one backbone with open release.

  31. Raon-OpenTTS: Open Models and Data for Robust Text-to-Speech

    eess.AS 2026-05 unverdicted novelty 5.0

    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.

  32. JaiTTS: A Thai Voice Cloning Model

    cs.CL 2026-04 unverdicted novelty 5.0

    JaiTTS-v1.0 achieves 1.94% CER on short Thai speech, beating human ground truth of 1.98%, matches humans on long speech, and wins 283 of 400 human comparisons against commercial systems.

  33. JaiTTS: A Thai Voice Cloning Model

    cs.CL 2026-04 unverdicted novelty 5.0

    JaiTTS-v1.0 achieves a character error rate of 1.94% on short Thai speech tasks, surpassing human ground truth of 1.98%, matches humans on long tasks, and wins 283 of 400 human pairwise comparisons against commercial models.

  34. Multimodal Large Language Model-Enabled Video Translation: A Role-Oriented Survey

    cs.CV 2026-04 accept novelty 5.0

    MLLM-enabled video translation is usefully framed as three roles—Semantic Reasoner, Expressive Performer, and Visual Synthesizer—rather than a cascade of ASR, MT, TTS, and lip-sync.

  35. Enhancing Speech Large Language Models through Reinforced Behavior Alignment

    cs.CL 2025-08 unverdicted novelty 5.0

    Reinforced Behavior Alignment (RBA) uses self-synthesized data from a teacher LLM and reinforcement learning to close the instruction-following gap in SpeechLMs, outperforming distillation and reaching SOTA on spoken ...

  36. Kimi-Audio Technical Report

    eess.AS 2025-04 unverdicted novelty 5.0

    Kimi-Audio is an open-source audio foundation model that achieves state-of-the-art results on speech recognition, audio understanding, question answering, and conversation after pre-training on more than 13 million ho...

  37. EntangleCodec: A Unified Discrete Audio Tokenizer via Semantic-Acoustic Entanglement

    cs.SD 2026-06 unverdicted novelty 4.0

    EntangleCodec unifies semantic and acoustic audio tokenization via caption alignment and flow-matching decoding, reporting competitive reconstruction, +7.4% gains on MMAR understanding, and 0.6B-parameter ALMs surpass...

  38. TRACE: Temporal Reasoning over Context and Evidence for Activity Recognition in Smart Homes

    cs.HC 2026-05 unverdicted novelty 4.0

    TRACE improves activity recognition accuracy and temporal coherence in smart homes by integrating multi-source sensor evidence with contextual priors.

  39. Multimodal Large Language Model-Enabled Video Translation: A Role-Oriented Survey

    cs.CV 2026-04 unverdicted novelty 4.0

    The paper offers the first focused review of MLLM-based video translation organized by a three-role taxonomy of Semantic Reasoner, Expressive Performer, and Visual Synthesizer, plus open challenges.

  40. AT-ADD: All-Type Audio Deepfake Detection Challenge Evaluation Plan

    cs.SD 2026-04 unverdicted novelty 3.0

    AT-ADD introduces standardized tracks and datasets for evaluating audio deepfake detectors on speech under real-world conditions and on diverse unknown audio types to promote generalization beyond speech-centric methods.