# FormalASR: End-to-End Spoken Chinese to Formal Text

arXiv: https://arxiv.org/abs/2605.19266 · HTML: https://mlsanity.com/paper/2605.19266/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 21 references that are in the map.
- First posted on arXiv 19 May 2026.

## Builds on (most cited first, 11 of 11)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale](https://mlsanity.com/paper/2208.07339/) (2022)
- [Listen, Attend and Spell](https://mlsanity.com/paper/1508.01211/) (2015)
- [WenetSpeech: A 10000+ Hours Multi-domain Mandarin Corpus for Speech Recognition](https://mlsanity.com/paper/2110.03370/) (2021)
- [FunAudioLLM: Voice Understanding and Generation Foundation Models for Natural Interaction Between Humans and LLMs](https://mlsanity.com/paper/2407.04051/) (2024)
- [Qwen3-ASR Technical Report](https://mlsanity.com/paper/2601.21337/) (2026)
- [HyPoradise: An Open Baseline for Generative Speech Recognition with Large Language Models](https://mlsanity.com/paper/2309.15701/) (2023)
- [Disfluency Detection using a Bidirectional LSTM](https://mlsanity.com/paper/1604.03209/) (2016)
- [RNN Approaches to Text Normalization: A Challenge](https://mlsanity.com/paper/1611.00068/) (2016)
- [Improving Disfluency Detection by Self-Training a Self-Attentive Model](https://mlsanity.com/paper/2004.05323/) (2020)

## Built on by (most cited first, 1 of 1)

- [AgenticASR: Refining Speech Recognition in Real-World Scenarios via an Agentic Approach](https://mlsanity.com/paper/2607.28175/) (2026)

## Abstract (arXiv)

Automatic speech recognition (ASR) systems are typically optimized for verbatim transcription, which preserves disfluencies, filler words, and informal spoken structures that are often unsuitable for downstream writing-oriented applications. A common workaround is a two-stage ASR+LLM pipeline for post-editing, but this design increases latency and memory cost and is difficult to deploy on-device. We present FormalASR, two compact end-to-end models (0.6B and 1.7B) that directly transcribe spoken Chinese into formal written text. To enable this setting, we build WenetSpeech-Formal and Speechio-Formal, two large-scale spoken-to-formal datasets constructed by LLM-based rewriting and quality filtering. We then fine-tune Qwen3-ASR at two scales (0.6B and 1.7B) with supervised fine-tuning. Experiments on WenetSpeech-Formal and Speechio-Formal show that FormalASR achieves up to 37.4% relative CER reduction over verbatim baselines, while also improving ROUGE-L and BERTScore. FormalASR requires no post-processing LLM at deployment time, providing a lightweight, on-device solution for spoken-to-formal transcription.