PeakEngine: A Deterministic On-the-Fly Pruning Neural Network Accelerator for Hearing Instruments

Zuzana Jelcicova, Evangelia Kasapaki, Oskar Andersson, Jens Sparso

Research output: Contribution to journalJournal articleResearchpeer-review

37 Downloads (Pure)


Recurrent neural networks (RNNs) are well-suited for sequential tasks such as speech enhancement (SE). However, their performance comes with high-computational complexity and latency. This impedes their deployment to battery-powered and resource-constrained hearing instruments (HIs) that need to operate for 16–18 h daily at only a few milliwatts (mW). In this article, we introduce PeakEngine, a configurable ASIC accelerator that decreases the amount of computation and memory accesses, and thus latency, in a gated recurrent unit (GRU) by means of adaptive inference. The reduction is achieved by on-the-fly pruning that selects the top K elements based on magnitudes of delta changes across timesteps from both input and hidden state sequences. Since K is constant, it results in a deterministic execution time. PeakEngine is synthesized in a 22-nm CMOS process, and the simulations show that it dissipates 11.83 μJ per inference for the baseline (unpruned) network and only 4.14–5.04 μJ for the pruned networks, with maximum acceptable degradation to no degradation in the improvement in audio quality and intelligibility. Moreover, the inference is on average sped up 2.2–2.97×, hence meeting the real-time requirements imposed by a HI application. To the best of our knowledge, PeakEngine is the first ASIC accelerator for deterministic and dynamic pruning in RNNs targeting HIs and SE.
Original languageEnglish
JournalIEEE Transactions on Very Large Scale Integration (VLSI) Systems
Issue number1
Pages (from-to)150 - 163
Publication statusPublished - 2023


  • Deterministic execution time
  • dynamic prun- ing, hardware accelerator
  • Hearing instruments (HIs)
  • Min-heap
  • Recurrent Neural Networks (RNNs)
  • Speech Enhancement (SE)
  • Top K


Dive into the research topics of 'PeakEngine: A Deterministic On-the-Fly Pruning Neural Network Accelerator for Hearing Instruments'. Together they form a unique fingerprint.

Cite this