five

Trelis/eval-nova-3-multimed-hard-20260408-1934

收藏
Hugging Face2026-04-08 更新2026-04-12 收录
下载链接:
https://hf-mirror.com/datasets/Trelis/eval-nova-3-multimed-hard-20260408-1934
下载链接
链接失效反馈
官方服务:
资源简介:
--- tags: - whisper - evaluation - speech - speech-to-text --- # Evaluation Results: nova-3 Evaluation results from Whisper model evaluation. ## Summary | Model | WER | CER | |-------|-----|-----| | [deepgram/nova-3](https://huggingface.co/deepgram/nova-3) | 12.03% | 6.88% | ## Source Data - **Evaluation Dataset:** [Trelis/multimed-hard](https://huggingface.co/datasets/Trelis/multimed-hard) - **Model Evaluated:** [deepgram/nova-3](https://huggingface.co/deepgram/nova-3) ## Columns | Column | Description | |--------|-------------| | `audio` | Audio sample (if available from source dataset) | | `reference` | Ground truth transcription | | `prediction` | Model prediction | | `wer` | Word Error Rate for this sample | | `cer` | Character Error Rate for this sample | | `entities` | Entity annotations from source dataset | | `entity_cer` | Per-sample entity CER (-1.0 if no entities) | ## Entity CER **Overall Entity CER:** 19.85% | Category | CER | |----------|-----| | anatomy | 26.81% | | biomarker | 18.99% | | condition | 15.45% | | drug | 0.00% | | organisation | 26.73% | | procedure | 21.89% | --- *Generated by [Trelis Studio](https://studio.trelis.com)*

标签: - Whisper - 模型评估 - 语音 - 语音转文字 # 评估结果:nova-3 本内容为Whisper模型的评估结果。 ## 摘要 | 模型 | 词错误率(Word Error Rate, WER) | 字符错误率(Character Error Rate, CER) | |-------|-----|-----| | [deepgram/nova-3](https://huggingface.co/deepgram/nova-3) | 12.03% | 6.88% | ## 源数据 - **评估数据集:** [Trelis/multimed-hard](https://huggingface.co/datasets/Trelis/multimed-hard) - **待评估模型:** [deepgram/nova-3](https://huggingface.co/deepgram/nova-3) ## 字段说明 | 字段名 | 描述 | |--------|-------------| | `audio` | 音频样本(若源数据集提供) | | `reference` | 标准转录文本(真实标注) | | `prediction` | 模型预测转录文本 | | `wer` | 该样本的词错误率 | | `cer` | 该样本的字符错误率 | | `entities` | 源数据集自带的实体标注 | | `entity_cer` | 单样本实体字符错误率(若无实体则为-1.0) | ## 实体字符错误率 **整体实体字符错误率:19.85%** | 类别 | CER | |----------|-----| | 解剖学 | 26.81% | | 生物标志物 | 18.99% | | 病症 | 15.45% | | 药物 | 0.00% | | 组织机构 | 26.73% | | 诊疗操作 | 21.89% | *由 [Trelis Studio](https://studio.trelis.com) 生成*
提供机构:
Trelis
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作