Language-family prompt tuning and a BPE-token-extended tokenizer improve Whisper's WER and inference speed on eight Indian languages, with 250 added tokens per language as the best configuration.
The dawn of the human-machine era: A forecast of new and emerging language technologies
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Enhancing Whisper's Accuracy and Speed for Indian Languages through Prompt-Tuning and Tokenization
Language-family prompt tuning and a BPE-token-extended tokenizer improve Whisper's WER and inference speed on eight Indian languages, with 250 added tokens per language as the best configuration.