Introducing the Hume ASR Entity Dataset
Blog post from Hume
The Hume ASR Entity Dataset is a purpose-built resource for improving automatic speech recognition of names, addresses, codes, dates, amounts, and other details that voice agents must capture accurately. It contains about 800 hours of scripted speech across eight languages, with more than 350,000 tagged entities spanning 28 categories, and is designed to address difficult real-world cases such as foreign-origin names in local speech, varied accents, and recordings made on everyday devices in homes and offices. Around 6,000 speakers from 49 countries contributed, and each recording includes written, spoken, and entity-tagged labels to support both speech recognition and text formatting, with quality checks applied to every clip. The dataset is available for licensing, with additional collection and language options, and is delivered with audio files, metadata, and documentation about speakers and recording conditions.
| Trend | Post Mentions | Total Month Mentions | Posts | Companies | MoM |
|---|---|---|---|---|---|
| Voice AI | 4 | 1,653 | 202 | 45 | -40% |
Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.