| Presentation # | 2 |
| Session: | ASR IV |
| Location: | Kallirhoe Hall |
| Session Time: | Friday, December 21, 13:30 - 15:30 |
| Presentation Time: | Friday, December 21, 13:30 - 15:30 |
| Presentation: |
Poster
|
| Topic: |
Speech recognition and synthesis: |
| Paper Title: |
TRANSLITERATION BASED APPROACHES TO IMPROVE CODE-SWITCHED SPEECH RECOGNITION PERFORMANCE |
| Authors: |
Jesse Emond, Bhuvana Ramabhadran, Brian Roark, Pedro Moreno, Min Ma, Google, United States |
| Abstract: |
Code-switching is a commonly occuring phenomenon in many multilingual communities, wherein a speaker switches between languages within a single utterance. Conventional Word Error Rate (WER) is not sufficient for measuring the performance of an Automated Speech Recognition (ASR) system on code-mixed languages due to ambiguities in transcription, misspellings and borrowing of words from two different writing systems. These rendering errors artificially inflate the WER of an ASR system and complicate its evaluation. Furthermore, these errors make it harder to accurately evaluate modeling errors originating from the code-switched language and acoustic models. In this work, we propose the use of a new metric, transliteration-optimized Word Error Rate (toWER) that smoothes out many of these irregularities by mapping all text to one writing system and demonstrate a correlation with the amount of code-switching present in a language. We also present a novel approach to acoustic and language modeling for bilingual code-switched indic languages using the same transliteration approach. We demonstrate the robustness and generality of our proposed approach on state-of-the-art Neural Network based acoustic and language models. We obtain significant gains in ASR performance of up to 10% relative on Google Voice Search and dictation traffic in several Indic languages. |