SpeechPrune: Context-aware Token Pruning for Speech Information Retrieval

While current Speech Large Language Models (Speech LLMs) excel at short-form tasks, they struggle with the computational and representational demands of longer audio clips. To advance the model's capabilities with long-form speech, we introduce Speech Information Retrieval (SIR), a long-context task for Speech LLMs, and present SPIRAL, a 1,012-sample benchmark testing models’ ability to extract critical details from long spoken inputs. To overcome the challenges of processing long speech sequences, we propose SpeechPrune, a training-free token pruning strategy that uses speech-text similarity and approximated attention scores to efficiently discard irrelevant tokens. In SPIRAL, SpeechPrune achieves accuracy improvements of 29% and up to 47% over the original model and the random pruning model at a pruning rate of 20%, respectively. SpeechPrune can maintain network performance even at a pruning level of 80%. This highlights the potential of token-level pruning for efficient and scalable speech understanding.