From Senones to Chenones: Tied Context-Dependent Graphemes for Hybrid Speech Recognition
This work provides an alternative to end-to-end ASR by enabling hybrid systems to drop reliance on phonetic knowledge, improving performance on tasks like proper noun recognition.
The paper tackles the assumption that hybrid ASR systems cannot effectively model graphemes in English by introducing tied context-dependent graphemes (chenones), achieving relative improvements of 4.5% to 11.1% over senone baselines on three datasets and competitive results with end-to-end approaches.
There is an implicit assumption that traditional hybrid approaches for automatic speech recognition (ASR) cannot directly model graphemes and need to rely on phonetic lexicons to get competitive performance, especially on English which has poor grapheme-phoneme correspondence. In this work, we show for the first time that, on English, hybrid ASR systems can in fact model graphemes effectively by leveraging tied context-dependent graphemes, i.e., chenones. Our chenone-based systems significantly outperform equivalent senone baselines by 4.5% to 11.1% relative on three different English datasets. Our results on Librispeech are state-of-the-art compared to other hybrid approaches and competitive with previously published end-to-end numbers. Further analysis shows that chenones can better utilize powerful acoustic models and large training data, and require context- and position-dependent modeling to work well. Chenone-based systems also outperform senone baselines on proper noun and rare word recognition, an area where the latter is traditionally thought to have an advantage. Our work provides an alternative for end-to-end ASR and establishes that hybrid systems can be improved by dropping the reliance on phonetic knowledge.