representation space gap

speech-text alignment (1) speech large language model (1) data-efficient training (1) sequence length inconsistency (1) end-to-end speech (1) large language model (1) speech translation (1) multimodal learning (1) speech recognition (1) efficient training (1)