---
license: cc-by-nc-4.0
configs:
- config_name: default
  data_files:
  - split: eval
    path: data/eval-*
- config_name: transcripts
  data_files:
  - split: eval
    path: transcripts/eval-*
dataset_info:
- config_name: default
  features:
  - name: id
    dtype: string
  - name: mix
    dtype:
      audio:
        decode: false
  - name: transcript
    dtype: string
  - name: language
    dtype: string
  - name: speech_source
    dtype: string
  splits:
  - name: eval
    num_examples: 10
- config_name: transcripts
  features:
  - name: id
    dtype: string
  - name: variant
    dtype: string
  - name: human
    dtype: string
  - name: parakeet_nvidia_parakeet_tdt_0_6b_v3_20260814
    dtype: string
  - name: parakeet_cpp_lib_tdt_0_6b_v3_f16_20260814
    dtype: string
  - name: qwen_Qwen_Qwen3_ASR_0_6B_20260814
    dtype: string
  - name: whisper_turbo_turbo_20260814
    dtype: string
  splits:
  - name: eval
    num_bytes: 9410
    num_examples: 20
  download_size: 13149
  dataset_size: 9410
---

# Voice Focus Examples
Collection of examples for accurate foreground speaker transcription.

## Details

- **Curated by:** Joschka Wohlgemuth
- **Funded by:** ai-coustics GmbH

## Contact:
Web: https://ai-coustics.com