FreyaTTS: A Compact Tokenizer-Free Flow-Matching Transformer for Turkish-First Speech Synthesis

2026-07-10 · Ahmet Erdem Pamuk, Ömer Yentür, Ahmet Tunga Bayrak, Yavuz Alp Sencer Öztürk, Mustafa Yavuz arxiv

We introduce Freya-TTS, a compact, tokenizer-free, Turkish-first text-to-speech model designed for highly reliable and efficient conversational synthesis. Freya-TTS is a 183.2M-parameter non-autoregressive conditional flow-matching Diffusion Transformer (DiT) that operates in the continuous latent space of the frozen AudioVAE2 (16 kHz encode, 48 kHz decode), allowing the model to focus its capacity on text-to-latent mapping while inheriting high-quality 48 kHz reconstruction. We advance the framework along three key dimensions: (1) rule-free end-to-end modeling from a 92-symbol Turkish character vocabulary without a phonemizer, grapheme-to-phoneme frontend, or discrete speech tokenizer, with digit strings expanded to their spoken form at the text frontend; (2) non-autoregressive parallel denoising, which predicts the entire latent sequence simultaneously over a predicted duration; and (3) a production-oriented two-stage post-training recipe consisting of single-speaker voice locking and short-utterance coverage, improving speaker consistency and robustness on short inputs. On the Freya-TR-Eval benchmark, Freya-TTS achieves a band-matched word error rate (WER) of 8.0% and character error rate (CER) of 3.0%, lower error than both larger open systems in its field, XTTS-v2 and F5-TTS, at 40-55% of their parameter count, together with the highest naturalness (MOS) among the compact systems. The model achieves a real-time factor of 0.11 on a consumer GPU (RTX 4090; ~0.14 mean on an H100) and synthesizes in real time on a laptop CPU, making it well suited for resource-constrained edge deployment. We release the model weights, training and inference code, and evaluation benchmark under the Apache-2.0 license.

📄 PDF Abstract BibTeX arXiv:2607.09530

Code (0)

등록된 구현이 없습니다.

Tasks

Speech Synthesis

Yeti: A compact protein structure tokenizer for reconstruction and multi-modal generation

2026-05-11 · Nabin Giri, Steven Farrell, Kristofer E. Bouchard arxiv

Multimodal models that jointly reason over protein sequences, structures, and function annotations within a unified representation hold immense potential for integrating multimodal data and generating new proteins with d…

RecTok: Reconstruction Distillation along Rectified Flow

2025-12-15 · Qingyu Shi, Size Wu, Jinbin Bai, Kaidong Yu 외 arxiv

Visual tokenizers play a crucial role in diffusion models. The dimensionality of latent space governs both reconstruction fidelity and the semantic expressiveness of the latent feature. However, a fundamental trade-off i…

Image Reconstruction

Flow Autoencoders are Effective Protein Tokenizers

2025-09-30 · Rohit Dilip, Evan Zhang, Ayush Varshney, David Van Valen arxiv

Protein structure tokenizers enable the creation of multimodal models of protein structure, sequence, and function. Current approaches to protein structure tokenization rely on bespoke components that are invariant to sp…

HiFlow: Tokenization-Free Scale-Wise Autoregressive Policy Learning via Flow Matching

2026-03-28 · Daichi Yashima, Koki Seno, Shuhei Kurita, Yusuke Oda 외 arxiv

Coarse-to-fine autoregressive modeling has recently shown strong promise for visuomotor policy learning, combining the inference efficiency of autoregressive methods with the global trajectory coherence of diffusion-base…

Image Generation

Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and Editing

2026-07-21 · Xinjie Zhang, Peng Zhang, Shicheng Zheng, Jinghao Guo 외 arxiv

Large-scale visual generators are increasingly capable but costly to train, fine-tune, and deploy. We introduce Mage-Flow, a compact 4B-scale generative stack for efficient text-to-image generation and instruction-based …

Text-to-Image Generation Image Editing