Pith. sign in

REVIEW 14 cited by

BASE TTS: Lessons from building a billion-parameter Text-to-Speech model on 100K hours of data

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 2402.08093 v2 pith:WIUPLPKW submitted 2024-02-12 cs.LG cs.CLeess.AS

classification cs.LGcs.CLeess.AS
keywords basemodeltext-to-speechtextbfabilitiesdatahoursspeech
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

We introduce a text-to-speech (TTS) model called BASE TTS, which stands for $\textbf{B}$ig $\textbf{A}$daptive $\textbf{S}$treamable TTS with $\textbf{E}$mergent abilities. BASE TTS is the largest TTS model to-date, trained on 100K hours of public domain speech data, achieving a new state-of-the-art in speech naturalness. It deploys a 1-billion-parameter autoregressive Transformer that converts raw texts into discrete codes ("speechcodes") followed by a convolution-based decoder which converts these speechcodes into waveforms in an incremental, streamable manner. Further, our speechcodes are built using a novel speech tokenization technique that features speaker ID disentanglement and compression with byte-pair encoding. Echoing the widely-reported "emergent abilities" of large language models when trained on increasing volume of data, we show that BASE TTS variants built with 10K+ hours and 500M+ parameters begin to demonstrate natural prosody on textually complex sentences. We design and share a specialized dataset to measure these emergent abilities for text-to-speech. We showcase state-of-the-art naturalness of BASE TTS by evaluating against baselines that include publicly available large-scale text-to-speech systems: YourTTS, Bark and TortoiseTTS. Audio samples generated by the model can be heard at https://amazon-ltts-paper.com/.

Discussion (0). Sign in to comment.

Forward citations

Cited by 14 Pith papers

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

  1. The NaijaVoices Dataset: Cultivating Large-Scale, High-Quality, Culturally-Rich Speech Data for African Languages

    cs.CL 2025-05 conditional novelty 7.0 of 10

    NaijaVoices is a 1,838-hour, 5,455-speaker speech-text corpus for Igbo, Hausa, and Yoruba whose use in fine-tuning cuts Word Error Rates by 42-76% relative to unadapted baselines.

  2. MPO: Multidimensional Preference Optimization for Language Model-based Text-to-Speech

    eess.AS 2025-08 conditional novelty 6.0 of 10

    MPO improves TTS alignment by constructing multi-dimensional preference pairs and adding cross-entropy regularization to DPO, yielding better intelligibility, speaker similarity, and prosody.

  3. Robust and Efficient Autoregressive Speech Synthesis with Dynamic Chunk-wise Prediction Policy

    cs.SD 2025-06 conditional novelty 6.0 of 10

    DCAR dynamically schedules chunk-wise token prediction in AR TTS, improving WER by up to 72.27% relative and speeding up inference by up to 2.89x over next-token baselines.

  4. Dataset of News Articles with Provenance Metadata for Media Relevance Assessment

    cs.CL 2025-06 conditional novelty 6.0 of 10

    A new benchmark dataset and two tasks let researchers test whether AI systems can judge if a news image's recorded location and date match the article, with current chatbots scoring 64-81% on location but 42-58% on date.

  5. EmergentTTS-Eval: Evaluating TTS Models on Complex Prosodic, Expressiveness, and Linguistic Challenges Using Model-as-a-Judge

    cs.LG 2025-05 conditional novelty 6.0 of 10

    A new TTS benchmark with 1,645 LLM-generated challenge prompts and an audio-language-model judge that ranks TTS systems in close agreement with a small human study.

  6. VoiceStar: Robust Zero-Shot Autoregressive TTS with Duration Control and Extrapolation

    eess.AS 2025-05 conditional novelty 6.0 of 10

    VoiceStar uses a progress-based rotary position embedding and mixed prompt training to give zero-shot voice cloning precise duration control and much longer output than training clips.

  7. Eta-WavLM: Efficient Speaker Identity Removal in Self-Supervised Speech Representations Using a Simple Linear Equation

    cs.SD 2025-05 conditional novelty 6.0 of 10

    A linear residual that subtracts a speaker-embedding-based projection from WavLM representations reduces speaker information while preserving content, improving voice conversion.

  8. Quantize More, Lose Less: Autoregressive Generation from Residually Quantized Speech Representations

    cs.SD 2025-07 reject novelty 5.0 of 10

    QTTS models speech as sequences from a multi-codebook RVQ audio codec whose first codebook is trained with ASR supervision, aiming for higher-fidelity TTS than single-codebook systems.

  9. StreamFlow: Streaming Flow Matching with Block-wise Guided Attention Mask for Speech Token Decoding

    cs.SD 2025-06 conditional novelty 5.0 of 10

    StreamFlow achieves streaming speech token decoding with 180 ms first-packet latency by using hierarchical block-wise attention masks in a DiT flow matching model.

  10. Accelerating Flow-Matching-Based Text-to-Speech via Empirically Pruned Step Sampling

    eess.AS 2025-05 conditional novelty 5.0 of 10

    A hand-designed, training-free step schedule prunes later sampling steps in flow-matching TTS, cutting F5-TTS inference cost by about 4x while keeping quality roughly unchanged.

  11. CloneShield: A Framework for Universal Perturbation Against Zero-Shot Voice Cloning

    cs.SD 2025-05 reject novelty 5.0 of 10

    A universal adversarial perturbation framework claiming to protect speech against zero-shot voice cloning by degrading cloned outputs while preserving input naturalness.

  12. Discrete Audio Representations for Automated Audio Captioning

    cs.SD 2025-05 conditional novelty 5.0 of 10

    A supervised audio tokenizer trained with audio tagging labels nearly matches continuous BEATs features for automated audio captioning on Clotho, while conventional discrete tokens degrade performance.

  13. Accelerating Diffusion-based Text-to-Speech Model Training with Dual Modality Alignment

    eess.AS 2025-05 conditional novelty 4.0 of 10

    A-DMA adds text and speech alignment losses to F5-TTS training, cutting the number of steps to a target quality while also improving WER and speaker similarity.

  14. Accelerating Autoregressive Speech Synthesis Inference With Speech Speculative Decoding

    cs.SD 2025-05 conditional novelty 4.0 of 10

    A speculative decoding variant with a tolerance-based acceptance rule speeds up CosyVoice 2 inference by 1.4x while keeping subjective quality comparable.

Pith tools