# Pre-trained Speech Representations as Feature Extractors for Speech Quality Assessment in Online Conferencing Applications

arXiv: https://arxiv.org/abs/2210.00259 · HTML: https://mlsanity.com/paper/2210.00259/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 1 Oct 2022.

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

- [Non-Intrusive Speech Intelligibility Prediction for Hearing-Impaired Users using Intermediate ASR Features and Human Memory Models](https://mlsanity.com/paper/2401.13611/) (2024)
- [Perceive and predict: self-supervised speech representation based loss functions for speech enhancement](https://mlsanity.com/paper/2301.04388/) (2023)
- [On the Use of Self-Supervised Speech Representations in Spontaneous Speech Synthesis](https://mlsanity.com/paper/2307.05132/) (2023)
- [Non Intrusive Intelligibility Predictor for Hearing Impaired Individuals using Self Supervised Speech Representations](https://mlsanity.com/paper/2307.13423/) (2023)
- [Efficient Speech Quality Assessment using Self-supervised Framewise Embeddings](https://mlsanity.com/paper/2211.06646/) (2022)
- [SVSNet+: Enhancing Speaker Voice Similarity Assessment Models with Representations from Speech Foundation Models](https://mlsanity.com/paper/2406.08445/) (2024)
- [Bottleneck Transformer-Based Approach for Improved Automatic STOI Score Prediction](https://mlsanity.com/paper/2602.15484/) (2026)

## Abstract (arXiv)

Speech quality in online conferencing applications is typically assessed through human judgements in the form of the mean opinion score (MOS) metric. Since such a labor-intensive approach is not feasible for large-scale speech quality assessments in most settings, the focus has shifted towards automated MOS prediction through end-to-end training of deep neural networks (DNN). Instead of training a network from scratch, we propose to leverage the speech representations from the pre-trained wav2vec-based XLS-R model. However, the number of parameters of such a model exceeds task-specific DNNs by several orders of magnitude, which poses a challenge for resulting fine-tuning procedures on smaller datasets. Therefore, we opt to use pre-trained speech representations from XLS-R in a feature extraction rather than a fine-tuning setting, thereby significantly reducing the number of trainable model parameters. We compare our proposed XLS-R-based feature extractor to a Mel-frequency cepstral coefficient (MFCC)-based one, and experiment with various combinations of bidirectional long short term memory (Bi-LSTM) and attention pooling feedforward (AttPoolFF) networks trained on the output of the feature extractors. We demonstrate the increased performance of pre-trained XLS-R embeddings in terms a reduced root mean squared error (RMSE) on the ConferencingSpeech 2022 MOS prediction task.