metadata
language:
- de
license: apache-2.0
tags:
- voice
- classification
- age
- gender
- speech
- audio
datasets:
- mozilla-foundation/common_voice_12_0
widget:
- src: >-
https://huggingface.co/padmalcom/wav2vec2-asr-ultimate-german/resolve/main/test.wav
example_title: Sample 1
pipeline_tag: audio-classification
metrics:
- accuracy
This multi-task wav2vec2 based asr model has two additional classification heads to detect:
- age
- gender
... of the current speaker in one forward pass.
It was trained on mozilla common voice.
Code for training can be found here.
inference_online.py shows, how the model can be used.