Hugging Face's logo

Spaces:
Thanh-Lam
/
vietnamese-speaker-profiling-v2
Running

App Files Community
vietnamese-speaker-profiling-v2

Commit History

Enhance model loading: auto-detect head_hidden_dim from checkpoint and streamline checkpoint loading process
b0cfc60
Running

Thanh-Lam commited on

Update model files: modify model.safetensors and training_args.bin, and ensure newline at end of preprocessor_config.json
a74861c

Thanh-Lam commited on

Fix: correct gender/dialect label mapping (Female=0, Male=1) and remove trim-causing sources param
ec8293e

Thanh-Lam commited on

Fix: model returns dict, not tuple
8f4a2bc

Thanh-Lam commited on

Use librosa instead of torchaudio for audio loading
2c4c515

Thanh-Lam commited on

Remove all emoji/icons from UI
3799c1c

Thanh-Lam commited on

Fix: use correct parameter names num_genders and num_dialects
4b131eb

Thanh-Lam commited on

Fix: use MultiTaskSpeakerModel and remove theme parameter for Gradio compatibility
3c8c19b

Thanh-Lam commited on

Add multi-model support with PhoWhisper and model selection dropdown
f9308ba

Thanh-Lam commited on

Vietnamese Speaker Profiling with wav2vec2-base-vi-vlsp2020
c3418e9

Thanh-Lam commited on