# Learning To Detect Keyword Parts And Whole By Smoothed Max Pooling

arXiv: https://arxiv.org/abs/2001.09246 · HTML: https://mlsanity.com/paper/2001.09246/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 25 Jan 2020.

## Built on by (most cited first, 10 of 10)

- [Deep Spoken Keyword Spotting: An Overview](https://mlsanity.com/paper/2111.10592/) (2021)
- [Noisy student-teacher training for robust keyword spotting](https://mlsanity.com/paper/2106.01604/) (2021)
- [Production federated keyword spotting via distillation, filtering, and joint federated-centralized training](https://mlsanity.com/paper/2204.06322/) (2022)
- [Self-Learning for Personalized Keyword Spotting on Ultra-Low-Power Audio Sensors](https://mlsanity.com/paper/2408.12481/) (2024)
- [SpeakerStew: Scaling to Many Languages with a Triaged Multilingual Text-Dependent and Text-Independent Speaker Verification System](https://mlsanity.com/paper/2104.02125/) (2021)
- [Utilizing TTS Synthesized Data for Efficient Development of Keyword Spotting Model](https://mlsanity.com/paper/2407.18879/) (2024)
- [Locale Encoding For Scalable Multilingual Keyword Spotting Models](https://mlsanity.com/paper/2302.12961/) (2023)
- [Exploring Sequence-to-Sequence Transformer-Transducer Models for Keyword Spotting](https://mlsanity.com/paper/2211.06478/) (2022)
- [Personalizing Keyword Spotting with Speaker Information](https://mlsanity.com/paper/2311.03419/) (2023)
- [Adversarial training of Keyword Spotting to Minimize TTS Data Overfitting](https://mlsanity.com/paper/2408.10463/) (2024)

## Abstract (arXiv)

We propose smoothed max pooling loss and its application to keyword spotting systems. The proposed approach jointly trains an encoder (to detect keyword parts) and a decoder (to detect whole keyword) in a semi-supervised manner. The proposed new loss function allows training a model to detect parts and whole of a keyword, without strictly depending on frame-level labeling from LVCSR (Large vocabulary continuous speech recognition), making further optimization possible. The proposed system outperforms the baseline keyword spotting model in [1] due to increased optimizability. Further, it can be more easily adapted for on-device learning applications due to reduced dependency on LVCSR.