padmalcom's picture
Update README.md
10c0936
|
raw
history blame
732 Bytes
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.