Skip to content

Latest commit

 

History

History
52 lines (37 loc) · 1.2 KB

readme.md

File metadata and controls

52 lines (37 loc) · 1.2 KB
Raw

Downsampled Acoustic Word Embeddings

Overview

MFCCs are downsampled to obtain acoustic word embeddings. These are evaluated using same-different evaluation.

Downsampling

Perform downsampling on MFCCs without deltas:

mkdir -p exp/SP
./downsample.py --technique resample --frame_dims 13 \
    ../features/mfcc/CH/ch.eval.gt_words.npz \
    exp/CH/mfcc.eval.gt_words.downsample_10.npz 10

Evaluation

Evaluate and analyse downsampled MFCCs without deltas:

../embeddings/eval_samediff.py --mvn \
    exp/SP/mfcc.dev.gt_words.downsample_10.npz
../embeddings/analyse_embeds.py --normalize --word_type \
    guatemala,presidente,autoridades,candidatos,asesinato,presupuesto,vicepresidente,negociaciones,netanyahu,social,explotaciones \
    exp/SP/mfcc.dev.gt_words.downsample_10.npz

Results

SWDP average precision:

  • SP dev: 0.14567458

(Deprecated) SWDP average precision:

  • CH dev: 0.11420457

  • CR dev: 0.11620668

  • HA dev: 0.11831970

  • SP dev: 0.12301926

  • SW dev: 0.06808896

  • TU dev: 0.13914600

  • GE dev: 0.08031011

  • KO dev: 0.13563458

  • TH dev: 0.08781202

  • VN dev: 0.02734849

  • SP eval: 0.19438775