LAST: Language Model Aware Speech Tokenization
Open paper
Benchmark
Model
Rank
Results
language-modelling-on-salmon
LAST 1.3B
–
Sentiment Consistency: 65.0
Speaker Consistency: 64.5
…
language-modelling-on-salmon
LAST 350M
–
Sentiment Consistency: 64.0
Speaker Consistency: 63.0
…
Rank counts only results with a code link.