# End-to-End ASR-free Keyword Search from Speech

## TL;DR

This paper presents an end-to-end, ASR-free keyword search system from speech that uses neural networks to directly determine the presence of a text query in audio without relying on traditional speech recognition, enabling faster training.

## Contribution

It introduces a novel E2E KWS system combining acoustic auto-encoders and language models trained with minimal supervision, bypassing conventional ASR dependencies.

## Key findings

- Performs competitively without traditional ASR.
- Trains significantly faster than conventional systems.
- Effective in detecting keywords directly from speech.

## Abstract

End-to-end (E2E) systems have achieved competitive results compared to conventional hybrid hidden Markov model (HMM)-deep neural network based automatic speech recognition (ASR) systems. Such E2E systems are attractive due to the lack of dependence on alignments between input acoustic and output grapheme or HMM state sequence during training. This paper explores the design of an ASR-free end-to-end system for text query-based keyword search (KWS) from speech trained with minimal supervision. Our E2E KWS system consists of three sub-systems. The first sub-system is a recurrent neural network (RNN)-based acoustic auto-encoder trained to reconstruct the audio through a finite-dimensional representation. The second sub-system is a character-level RNN language model using embeddings learned from a convolutional neural network. Since the acoustic and text query embeddings occupy different representation spaces, they are input to a third feed-forward neural network that predicts whether the query occurs in the acoustic utterance or not. This E2E ASR-free KWS system performs respectably despite lacking a conventional ASR system and trains much faster.