First open Basque speech language model built on LLaMA-Omni
Fine-tuning a Latxa backbone on translated, synthesized VoiceAssistant-400K data scored 3.86/5 content quality and 9.94% WER.
Stories per day · 30 days
Relatedtext-to-speech2basque1cross-lingual1latxa1llama-omni1speech-language-model1speech-synthesis1vocabulary-expansion1
6 stories, newest first · one card per story
Stories tagged #low-resource
SEAM expands TTS vocabularies without paired audio fine-tuning
The method matches or approaches baselines with up to four times less data across three languages and two architectures.
Multilingual GSM-Symbolic: What determines capability transfer across languages?
Multilingual GSM-Symbolic finds model size, resources, reasoning, and typology drive cross-lingual math transfer.
Right Tool, Right Job: Native-Language Evaluation, Tokenizer Sensitivity, and Methodological Findings from a French-Only BabyLM
French BabyLM entry METRON-FR (125M GPT-2, 92.47M words) shows tokenizer artifacts dominate child-scale zero-shot evaluation; proposes standard diagnostics.
How Far Can Synthetic Data Take Thai OCR?
Synthetic-only training adapts PaddleOCR-VL into Wayu-Paxa-OCR-Zero, cutting Thai printed-page CER from 6.64% to 1.24% without real labels.
Building and Evaluating Fixed-Voice Thai TTS from Synthetic Speech
Researchers distill a compact 82M-parameter Thai TTS from synthetic OmniVoice data, enabling on-device fixed-voice synthesis without reference audio.