A unified phoneme-based TTS-to-ASR augmentation pipeline built around a multilingual TTS model trained from scratch using the F5-TTS architecture with language-ID conditioning is presented and phoneme-frequency-guided selection (PFGS) is proposed, which ranks candidate sentences using phoneme frequencies estimated from real ASR training labels.
Abstract
Synthetic speech offers scalable supervision for automatic speech recognition (ASR), but its benefit depends on text selection, reference speech, and augmentation scale. We present a phoneme-based TTS-to-ASR pipeline using a single TTS model jointly trained from scratch on Arabic, French, Italian, and Portuguese with the F5-TTS architecture and language-monolingual ASR systems cover 13 test sets. Across the synthesis-scale sweep, random augmentation improves over matched real-only continuation on 11 sets. In the selection comparison, PFGS improves over real-only training on 12 sets and over random selection on nine, with a maximum relative WER reduction of 19.3% against random selection. With target texts and synthesis counts fixed, reference-speech filtering reduces absolute WER by 0.29 and 0.59 points on Italian and French Common Voice, respectively. These findings support treating TTS augmentation as a synthetic-corpus construction problem, rather than merely a question of generation scale.
End-to-end on-device Automatic Speech Recognition (ASR) systems have demonstrated remarkable accuracy and efficiency in recent years. However, challenges persist in correctly transcribing infrequent named entities (e.g., geographical locations, business entities, person names, etc.) and handling diverse user accents, which remain underrepresented in training datasets. While information retrieval augmentation or Retrieval Augmentation Generation (RAG) for correction of named entities has shown promise in knowledge-grounded NLP tasks when paired with large language models (LLMs), its integration into real-time on-device systems is non-trivial due to computational constraints. We introduce a novel lightweight method combining phonetic-aware retrieval, vector-based semantic search and generative correction. The system leverages a lightweight phonetic index for rapid candidate entity retrieval and a dense vectorembedding module to refine predictions as well as model the ASR error output distribution in generative space. Additionally, we introduce a novel approach to model ASR errors in natural language. Experiments on test sets emphasizing place names, monuments, airports, and landscapes yielded an increase in correct Named Entity (NE) recognition accuracy by 9.6% compared to baseline. These gains underscore the efficacy of hybrid retrieval-generation paradigms in resource-constrained environments.
Kiranmayi Gandikota, Anunay Katare, C. Pandey et al.· International Conference on...· 0 citations
In low-resource settings, deploying TTS typically requires choosing between a large voice-cloning model with costly inference or a compact fixed-voice system that requires a speaker-specific corpus. We study a third route: using a large voice-cloning model as a programmable data source to turn a short voice reference (e.g., 15 seconds) into a compact fixed-voice student trained entirely on synthetic speech. This setting makes pipeline design consequential: teacher errors become training targets, while filtering failed generations can reduce coverage of difficult texts. Thai further introduces challenges from ambiguous word boundaries, lexical tone, names and loanwords, numeric verbalization, and Thai-English code-switching. We study how text preparation, synthetic generation, quality filtering, rejection sampling, and frontend choices affect the resulting student, and where teacher limitations remain. We evaluate CER, Challenge-Set Keyword Accuracy, Prosody Pause Accuracy, speaker similarity, and speaking rate. The resulting 82M-parameter model, Wayu-Paxa-TTS-Edge, enables on-device Thai TTS without reference audio. It achieves 68.2% Challenge-Set Keyword Accuracy (85.5% of Gemini 3.1) and 91.4% pause precision, outperforming its OmniVoice teacher (89.9%) and reaching 94.8% of Gemini 3.1. It also achieves the lowest pause-placement error and intra-word pause rates among the three systems, and 3.7% and 1.1% CER on Thai and English, respectively. We open-source the model and evaluation framework for Thai TTS development.
Kunat Pipatanakul, Potsawee Manakul, Warit Sirichotedumrong et al.· 0 citations
Current Large Language Models (LLMs) are primarily optimized for written text, often producing outputs that are grammatically correct and helpful yet poorly suited for spoken delivery via Text-to-Speech (TTS). In this work, we study how to make LLMs natively generate TTS-friendly text, which we frame as a preference alignment problem: instead of relying on downstream rewriting modules, we directly align LLMs to generate text optimized for spoken delivery. We introduce two preference datasets spanning different target domains, CORA and Recipe, which contain paired TTS-friendly and TTS-unfriendly responses. We further propose an evaluation suite combining a pattern-based heuristic metric, a TTS$\to$ASR evaluation pipeline, and a MUSHRA listening study with human judges. Our experiments compare the recently proposed Feature-aware Sampling and Tuning (FaST) framework -- leveraging interpretable features instead of a black-box reward model -- against an array of alignment baselines on the TTS-friendly generation task. Notably, we found that FaST achieves the best overall tradeoff between TTS-friendliness and helpfulness across various settings. We also identified a strong correlation between our different metrics, highlighting the ability to reliably assess TTS-friendliness via an efficient heuristic.
Thibaut Thonet, Jos Rozen, Laurent Besacier· 0 citations
Findings indicate that language-specific fine-tuning plays a more critical role than multilingual generalization in achieving accurate ASR for Indonesian and provide practical guidance for deploying ASR systems in low-resource language scenarios.
J. Hebert, Amalia Zahra· Bulletin of Electrical Engin...· 0 citations
Results show that supervised fine-tuning provides the largest gain, while synthetic-speech LoRA adaptation and reinforcement learning further improve robustness, while synthetic-speech LoRA adaptation and reinforcement learning further improve robustness.
Hao Wu, Rong-Qi Han, Zhen Wang et al.· 0 citations
We present DONDO, a family of open, permissively licensed automatic speech recognition (ASR) base models for African languages, built on the w2v-BERT 2.0 self-supervised speech encoder. DONDO comprises twenty-one monolingual models and five multilingual models spanning twenty-seven language varieties across Ghana, Sierra Leone, Nigeria, Senegal, Kenya and Zimbabwe. Models are fine-tuned primarily on read speech drawn from religious texts, which offer broad, license-clear and orthographically consistent coverage for languages that otherwise lack transcribed audio. We describe a two-step (and, for one family, three-step) learning-rate-annealed fine-tuning procedure that first adapts a shared multilingual model at a high learning rate and then anneals it to recover, and in several cases surpass, strong monolingual baselines. We further describe a lightweight language-conditioning mechanism that injects a one-hot language identity as a sequence of prefix frames prepended to the acoustic features, allowing a single multilingual checkpoint to be steered to a target language at inference. Across the five multilingual families the annealed models reach average word error rates (WER) of 10-13%, closing most of the gap to monolingual models while covering many languages in a single checkpoint. All models are released on the Hugging Face KhayaAI organisation under the Apache-2.0 license (attribution only) so that others may fine-tune them freely, including for commercial use. We provide a conservative estimate that the languages covered are spoken by on the order of one hundred million first-language speakers, and by substantially more when second-language use is included.
P. Azunre, N. Ibrahim, Joel Budu et al.· 0 citations
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduSep 2, 2026
A new method, called CW-Net, translates the reasoning process of an autonomous vehicle’s AI system into understandable concepts that explain its behavior.
MIT News · Artificial Intelligence· news.mit.eduAug 31, 2026
With millions of users across the world, Julia has been used to conduct cutting-edge research and to design new drugs, jet engines, heat pumps, and more.
A new machine-learning framework aims to improve the success rate of computational protein design while moving away from results that reproduce sequences found in nature.
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
