# Improving child speech recognition with augmented child-like speech

arXiv: https://arxiv.org/abs/2406.10284 · HTML: https://mlsanity.com/paper/2406.10284/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 15 of its 36 references that are in the map.
- First posted on arXiv 12 Jun 2024.

## Builds on (most cited first, 15 of 15)

- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Conformer: Convolution-augmented Transformer for Speech Recognition](https://mlsanity.com/paper/2005.08100/) (2020)
- [SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition](https://mlsanity.com/paper/1904.08779/) (2019)
- [ESPnet: End-to-End Speech Processing Toolkit](https://mlsanity.com/paper/1804.00015/) (2018)
- [Snapshot Ensembles: Train 1, get M for free](https://mlsanity.com/paper/1704.00109/) (2017)
- [Generalized End-to-End Loss for Speaker Verification](https://mlsanity.com/paper/1710.10467/) (2017)
- [End-to-End Speech Recognition: A Survey](https://mlsanity.com/paper/2303.03329/) (2023)
- [Voice Conversion Challenge 2020: Intra-lingual semi-parallel and cross-lingual voice conversion](https://mlsanity.com/paper/2008.12527/) (2020)
- [AGAIN-VC: A One-shot Voice Conversion using Activation Guidance and Adaptive Instance Normalization](https://mlsanity.com/paper/2011.00316/) (2020)
- [Making More of Little Data: Improving Low-Resource Automatic Speech Recognition Using Data Augmentation](https://mlsanity.com/paper/2305.10951/) (2023)
- [Voice Conversion Can Improve ASR in Very Low-Resource Settings](https://mlsanity.com/paper/2111.02674/) (2021)
- [Exploring data augmentation in bias mitigation against non-native-accented speech](https://mlsanity.com/paper/2312.15499/) (2023)
- [Data Augmentation For Children's Speech Recognition -- The "Ethiopian" System For The SLT 2021 Children Speech Recognition Challenge](https://mlsanity.com/paper/2011.04547/) (2020)
- [The SLT 2021 children speech recognition challenge: Open datasets, rules and baselines](https://mlsanity.com/paper/2011.06724/) (2020)
- [Spectral Modification Based Data Augmentation For Improving End-to-End ASR For Children's Speech](https://mlsanity.com/paper/2203.06600/) (2022)

## Built on by (most cited first, 7 of 7)

- [Speech Recognition for Automatically Assessing Afrikaans and isiXhosa Preschool Oral Narratives](https://mlsanity.com/paper/2501.06478/) (2025)
- [Towards Inclusive ASR: Investigating Voice Conversion for Dysarthric Speech Recognition in Low-Resource Languages](https://mlsanity.com/paper/2505.14874/) (2025)
- [Improving Child Speech Recognition and Reading Mistake Detection by Using Prompts](https://mlsanity.com/paper/2506.11079/) (2025)
- [Finding My Voice: Generative Reconstruction of Disordered Speech for Automated Clinical Evaluation](https://mlsanity.com/paper/2509.19231/) (2025)
- [Egocentric Speaker Classification in Child-Adult Dyadic Interactions: From Sensing to Computational Modeling](https://mlsanity.com/paper/2409.09340/) (2024)
- [Utterance-Level Methods for Identifying Reliable ASR-Output for Child Speech](https://mlsanity.com/paper/2604.19801/) (2026)
- [Transcribing Children's Speech: ASR Performance and Obtaining Reliable Orthographic Transcriptions](https://mlsanity.com/paper/2605.28833/) (2026)

## Abstract (arXiv)

State-of-the-art ASRs show suboptimal performance for child speech. The scarcity of child speech limits the development of child speech recognition (CSR). Therefore, we studied child-to-child voice conversion (VC) from existing child speakers in the dataset and additional (new) child speakers via monolingual and cross-lingual (Dutch-to-German) VC, respectively. The results showed that cross-lingual child-to-child VC significantly improved child ASR performance. Experiments on the impact of the quantity of child-to-child cross-lingual VC-generated data on fine-tuning (FT) ASR models gave the best results with two-fold augmentation for our FT-Conformer model and FT-Whisper model which reduced WERs with ~3% absolute compared to the baseline, and with six-fold augmentation for the model trained from scratch, which improved by an absolute 3.6% WER. Moreover, using a small amount of "high-quality" VC-generated data achieved similar results to those of our best-FT models.