Preprint
HyPoradise: An Open Baseline for Generative Speech Recognition with Large Language Models
Chen Chen et al., arXiv (NeurIPS 2023 dataset track), 2023
Read the originalRead on October 16, 2026
- What it is
- A benchmark paper that uses large language models to correct speech recognition output from N-best hypothesis lists. Its abstract reports that the language model can produce words that were missing from the recognizer's candidate list.
- How much weight it carries
- A preprint with a public dataset and code. It studies research-style error correction on benchmark data, not the commercial dictation apps, so it shows what a language model can do to recognizer output and not what any app does.
Facts taken from this source
- In the HyPoradise benchmark, a large language model can correct words that are missing from the recognizer's N-best list, because it generates text rather than only choosing among candidates.
“LLM with reasonable prompt and its generative capability can even correct those tokens that are missing in N-best list” (Abstract, final sentences)
- The HyPoradise benchmark contains more than 334,000 pairs of recognizer N-best hypotheses and their correct transcriptions, for training language models to correct speech recognition errors.
“encompassing more than 334,000 pairs of N-best hypotheses and corresponding accurate transcriptions” (Abstract)
- The HyPoradise benchmark for language-model correction of speech recognition contains more than 334,000 pairs of N-best hypotheses and accurate transcriptions.
“encompassing more than 334,000 pairs of N-best hypotheses and corresponding accurate transcriptions across prevalent speech domains” (Abstract)
Pages that rest on this source
- Dictation Software: How to Choose the Right AppTopic
- Speech to Text AI: Dictation, Assistants and Voice AgentsTopic
- AI Dictation: How AI Voice Typing WorksGuide
- How Speech Recognition Works, From Sound to TextGuide
- What Is Dictation Software? Types and How It WorksGuide
- Is Speech to Text AI? What Kind of AI Voice Typing UsesGuide
- AI Cleanup for Dictation: Fillers, Punctuation, RewritesGuide
- Speech to Text for AI: Why Spoken Prompts Work BetterGuide
- Speech to Text and LLMs: What Language Models ChangeGuide
- Speech Recognition Models: How Neural ASR WorksGuide
- AI Dictation Apps Compared: How the Major Apps DifferGuide
- Can large language models do speech recognition?Question
- Do AI chatbots understand prompts with dictation errors?Question
- Do LLMs make speech recognition more accurate?Question
- Is speech recognition part of NLP?Question
- Is speech to text generative AI?Question
- Is there a dictation app without AI?Question
- Is transcription software considered generative AI?Question
- What is the best voice to text app for journaling?Question
- What is the best voice to text app for notes?Question
- What is the difference between speech to text and AI?Question
- AI dictationGlossary
- Generative error correction (GER)Glossary
- Large language model (LLM)Glossary