# The Microsoft 2016 Conversational Speech Recognition System

arXiv: https://arxiv.org/abs/1609.03528 · HTML: https://mlsanity.com/paper/1609.03528/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 12 Sep 2016.

## Built on by (most cited first, 9 of 9)

- [A Comprehensive Survey on Hardware-Aware Neural Architecture Search](https://mlsanity.com/paper/2101.09336/) (2021)
- [Hamiltonian Deep Neural Networks Guaranteeing Non-vanishing Gradients by Design](https://mlsanity.com/paper/2105.13205/) (2021)
- [A unified framework for Hamiltonian deep neural networks](https://mlsanity.com/paper/2104.13166/) (2021)
- [Avoid Overfitting User Specific Information in Federated Keyword Spotting](https://mlsanity.com/paper/2206.08864/) (2022)
- [InferNet: Exploiting Aggregate GPU Profiles as Side-Channel for DNN Architecture Inference](https://mlsanity.com/paper/2304.03388/) (2023)
- [Towards efficient end-to-end speech recognition with biologically-inspired neural networks](https://mlsanity.com/paper/2110.02743/) (2021)
- [A Distributed Optimisation Framework Combining Natural Gradient with Hessian-Free for Discriminative Sequence Training](https://mlsanity.com/paper/2103.07554/) (2021)
- [Towards the Transferable Audio Adversarial Attack via Ensemble Methods](https://mlsanity.com/paper/2304.08811/) (2023)
- [A Treatise On FST Lattice Based MMI Training](https://mlsanity.com/paper/2210.08918/) (2022)

## Abstract (arXiv)

We describe Microsoft's conversational speech recognition system, in which we combine recent developments in neural-network-based acoustic and language modeling to advance the state of the art on the Switchboard recognition task. Inspired by machine learning ensemble techniques, the system uses a range of convolutional and recurrent neural networks. I-vector modeling and lattice-free MMI training provide significant gains for all acoustic model architectures. Language model rescoring with multiple forward and backward running RNNLMs, and word posterior-based system combination provide a 20% boost. The best single system uses a ResNet architecture acoustic model with RNNLM rescoring, and achieves a word error rate of 6.9% on the NIST 2000 Switchboard task. The combined system has an error rate of 6.2%, representing an improvement over previously reported results on this benchmark task.