---
title: "Shanghai Jiao Tong's READ Cuts Speech Recognition Errors 20% Without Reference Transcripts"
subtitle: "READ flips ASR evaluation on its head, scoring transcripts by how well a TTS model can reconstruct the original audio, no reference text needed"
url: "https://alphasignal.ai/news/shanghai-jiao-tong-s-read-cuts-speech-recognition-errors-20-without-reference"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "AlphaSignal"
company_url: "https://alphasignal.ai/company/alphasignal"
published: "2026-06-03T00:00:00.000Z"
updated: "2026-06-12T01:02:09.831Z"
type: "Paper"
topics:
  - "Audio"
  - "Benchmarks"
subtopics:
  - "Speech To Text"
  - "Text To Speech"
upvotes: 2944
source_url: "https://arxiv.org/abs/2606.04680"
comments: 0
truncated: true
---

# Shanghai Jiao Tong's READ Cuts Speech Recognition Errors 20% Without Reference Transcripts

> READ flips ASR evaluation on its head, scoring transcripts by how well a TTS model can reconstruct the original audio, no reference text needed

Every time you evaluate a speech recognition system, you almost certainly compare its output against a human-written reference transcript. That dependency is so deeply baked into the field that the dominant metric, **Word Error Rate (WER)**, is literally defined as the edit distance between a hypothesis and a reference. But reference transcripts are expensive to produce, unavailable in production, and can themselves introduce errors. A team from Shanghai Jiao Tong University's X-LANCE Lab just published a paper proposing a fundamentally different approach: score a transcript by how well it explains the original audio, with no reference in sight.

The method is called [**READ**](https://arxiv.org/abs/2606.04680) (Reference-free Hypothesis Evaluation with Acoustic Discrepancy). The core idea is elegant: if a transcription is correct, then a good text-to-speech model should be able to reproduce audio that closely matches what was actually recorded. If the transcript is wrong, the TTS model's predicted audio will diverge from the real signal. READ quantifies that divergence as a score.

## Why the old way keeps falling short

Traditionally, ASR evaluation compares system outputs with ground-truth transcripts. Reference-based metrics like WER are calculated by counting errors against those transcripts, but their most significant limitation is that they require ground-truth transcripts, which may not always be available, and a poor reference can affect evaluation accuracy.

Referenceless evaluation metrics for ASR do not rely on a ground-truth reference. Some existing approaches use confidence scores and confidence-based fusion, where confidence scores are generated by the ASR system itself and used to rank alternative hypotheses. The problem is that these scores are internal to the model that made the error in the first place. End-to-end models show increased robustness and accuracy but use differing confidence estimation techniques and exhibit overconfidence. Asking a model to grade its own homework is a structural flaw.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [Read What You Hear: Reference-Free Hypotheses Evaluation with Acoustic Discrepancy](https://arxiv.org/abs/2606.04680)
- [READ paper HTML full text](https://arxiv.org/html/2606.04680)
- [NoRefER: a Referenceless Quality Metric for ASR via Semi-Supervised Language Model Fine-Tuning](https://arxiv.org/pdf/2306.12577)
- [On the Robust Approximation of ASR Metrics](https://arxiv.org/pdf/2502.12408)
- [Evaluating ASR Confidence Scores for Automated Error Detection](https://arxiv.org/html/2503.15124v1)