It doesn't. It implements a new way of decoding the logits which improves performance by a relative 16% over the previously best German speech recognition, whic

It doesn't. It implements a new way of decoding the logits which improves performance by a relative 16% over the previously best German speech recognition, which was Facebook's wav2vec2. And the size is relevant to people in the industry because DeepSpeech uses 2000+ LOCs for implementing their decoding, so this works better and is 10x less code.