Marie Kunešová
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Marie Kunešová has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Zbyněk Zajíc
University of West Bohemia
6 shared publications - Jindřich Matoušek
University of West Bohemia in Pilsen
4 shared publications - Vlasta Radová
University of West Bohemia
4 shared publications - Marek Hrúz
University of West Bohemia
3 shared publications - Zdeněk Hanzlíček
University of West Bohemia in Pilsen
3 shared publications - Aleš Pražák
University of West Bohemia in Pilsen
1 shared publication
Latest Publications
- Open Access
- 1 October 2026
- Published by Taylor & Francis in Accountability in Research
- Vol. 33 (8)
Opening Pandora’s box: Paper mills in conference proceedings
Maintaining research integrity is crucial for the credibility of science. A key threat to research integrity is the rise of paper mills, commercial entities that sell co-authorship slots and/or vio...
- 1 October 2026
- Published by Taylor & Francis in Accountability in Research
- Vol. 33 (8)
Open AccessOpening Pandora’s box: Paper mills in conference proceedings
Maintaining research integrity is crucial for the credibility of science. A key threat to research integrity is the rise of paper mills, commercial entities that sell co-authorship slots and/or vio...
- 31 August 2026
- Published by Springer Nature
- p. 652-663
Comparing Acoustic and Linguistic Features for Automated Deception Detection
- 31 August 2026
- Published by Springer Nature
- p. 652-663
Comparing Acoustic and Linguistic Features for Automated Deception Detection
- 22 August 2025
- Published by Springer Nature
- p. 3-12
Lightweight Target-Speaker-Based Overlap Transcription for Practical Streaming ASR
- 22 August 2025
- Published by Springer Nature
- p. 3-12
Lightweight Target-Speaker-Based Overlap Transcription for Practical Streaming ASR
- 22 August 2025
- Published by Springer Nature
- p. 84-95
An Exploration of ECAPA-TDNN and x-vector Speaker Representations in Zero-Shot Multi-speaker TTS
- 22 August 2025
- Published by Springer Nature
- p. 84-95
An Exploration of ECAPA-TDNN and x-vector Speaker Representations in Zero-Shot Multi-speaker TTS
- Open Access
- 7 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 140152-140174
Three Years of VoiceMOS Challenges: Lessons Learned by the UWB-NTIS-TTS Team
Automatic prediction of mean-opinion scores (MOS) promises a faster, cheaper alternative to listening tests, yet robust generalization across speakers, languages, and domains remains a significant challenge. This article presents our system designs and experimental results from three years of participation in the VoiceMOS Challenges (2022–2024), covering MOS prediction for synthesized or voice-converted speech and singing voice, including out-of-domain and cross-language conditions. We evaluate six neural architectures – wav2vec 2.0, QuartzNet, CNN-RNN, LDNet, RawNet3, and HiFi-GAN – and their ensembles. Across all tasks, we find that 1) self-supervised acoustic encoders are the most consistently reliable foundation, 2) ensembling yields rapidly diminishing returns once complementary representations are covered, and 3) the diversity and balance of training data outweigh architectural complexity. Notably, the indiscriminate fusion strategy that performed well in 2022 degrades under the mismatched French TTS conditions of 2023, emphasizing the importance of out-of-domain validation. Further experiments show that carefully pruned ensembles can modestly outperform the best single model while remaining within real-time constraints. We conclude with several observations to guide the development of computationally efficient, domain-robust MOS prediction systems.
- 7 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 140152-140174
Open AccessThree Years of VoiceMOS Challenges: Lessons Learned by the UWB-NTIS-TTS Team
Automatic prediction of mean-opinion scores (MOS) promises a faster, cheaper alternative to listening tests, yet robust generalization across speakers, languages, and domains remains a significant challenge. This article presents our system designs and experimental results from three years of participation in the VoiceMOS Challenges (2022–2024), covering MOS prediction for synthesized or voice-converted speech and singing voice, including out-of-domain and cross-language conditions. We evaluate six neural architectures – wav2vec 2.0, QuartzNet, CNN-RNN, LDNet, RawNet3, and HiFi-GAN – and their ensembles. Across all tasks, we find that 1) self-supervised acoustic encoders are the most consistently reliable foundation, 2) ensembling yields rapidly diminishing returns once complementary representations are covered, and 3) the diversity and balance of training data outweigh architectural complexity. Notably, the indiscriminate fusion strategy that performed well in 2022 degrades under the mismatched French TTS conditions of 2023, emphasizing the importance of out-of-domain validation. Further experiments show that carefully pruned ensembles can modestly outperform the best single model while remaining within real-time constraints. We conclude with several observations to guide the development of computationally efficient, domain-robust MOS prediction systems.
- Open Access
An Exploration of ECAPA-TDNN and x-vector Speaker Representations in Zero-shot Multi-speaker TTS
Zero-shot multi-speaker text-to-speech (TTS) systems rely on speaker embeddings to synthesize speech in the voice of an unseen speaker, using only a short reference utterance. While many speaker embeddings have been developed for speaker recognition, their relative effectiveness in zero-shot TTS remains underexplored. In this work, we employ a YourTTS-based TTS system to compare three different speaker encoders - YourTTS's original H/ASP encoder, x-vector embeddings, and ECAPA-TDNN embeddings - within an otherwise fixed zero-shot TTS framework. All models were trained on the same dataset of Czech read speech and evaluated on 24 out-of-domain target speakers using both subjective and objective methods. The subjective evaluation was conducted via a listening test focused on speaker similarity, while the objective evaluation measured cosine distances between speaker embeddings extracted from synthesized and real utterances. Across both evaluations, the original H/ASP encoder consistently outperformed the alternatives, with ECAPA-TDNN showing better results than x-vectors. These findings suggest that, despite the popularity of ECAPA-TDNN in speaker recognition, it does not necessarily offer improvements for speaker similarity in zero-shot TTS in this configuration. Our study highlights the importance of empirical evaluation when reusing speaker recognition embeddings in TTS and provides a framework for additional future comparisons.
Open AccessAn Exploration of ECAPA-TDNN and x-vector Speaker Representations in Zero-shot Multi-speaker TTS
Zero-shot multi-speaker text-to-speech (TTS) systems rely on speaker embeddings to synthesize speech in the voice of an unseen speaker, using only a short reference utterance. While many speaker embeddings have been developed for speaker recognition, their relative effectiveness in zero-shot TTS remains underexplored. In this work, we employ a YourTTS-based TTS system to compare three different speaker encoders - YourTTS's original H/ASP encoder, x-vector embeddings, and ECAPA-TDNN embeddings - within an otherwise fixed zero-shot TTS framework. All models were trained on the same dataset of Czech read speech and evaluated on 24 out-of-domain target speakers using both subjective and objective methods. The subjective evaluation was conducted via a listening test focused on speaker similarity, while the objective evaluation measured cosine distances between speaker embeddings extracted from synthesized and real utterances. Across both evaluations, the original H/ASP encoder consistently outperformed the alternatives, with ECAPA-TDNN showing better results than x-vectors. These findings suggest that, despite the popularity of ECAPA-TDNN in speaker recognition, it does not necessarily offer improvements for speaker similarity in zero-shot TTS in this configuration. Our study highlights the importance of empirical evaluation when reusing speaker recognition embeddings in TTS and provides a framework for additional future comparisons.
- Open Access
- 10 October 2024
- Published by Springer Nature in International Journal of Speech Technology
- Vol. 27 (4) , 847-859
Comparison of wav2vec 2.0 models on three speech processing tasks
The current state-of-the-art for various speech processing problems is a sequence-to-sequence model based on a self-attention mechanism known as transformer. The widely used wav2vec 2.0 is a self-supervised transformer model pre-trained on large amounts of unlabeled speech and then fine-tuned for a specific task. The data used for training and fine-tuning, along with the size of the transformer model, play a crucial role in both of these training steps. The most commonly used wav2vec 2.0 models are trained on relatively “clean” data from sources such as the LibriSpeech dataset, but we can expect there to be a benefit in using more realistic data gathered from a variety of acoustic conditions. However, it is not entirely clear how big the difference would be. Investigating this is the main goal of our article. To this end, we utilize wav2vec 2.0 models in three fundamental speech processing tasks: speaker change detection, voice activity detection, and overlapped speech detection, and test them on four real conversation datasets. We compare four wav2vec 2.0 models with different sizes and different data used for pre-training, and we fine-tune them either on in-domain data from the same dataset or on artificial training data created from the LibriSpeech corpus. Our results suggest that richer data that are more similar to the task domain bring better performance than a larger model.
- 10 October 2024
- Published by Springer Nature in International Journal of Speech Technology
- Vol. 27 (4) , 847-859
Open AccessComparison of wav2vec 2.0 models on three speech processing tasks
The current state-of-the-art for various speech processing problems is a sequence-to-sequence model based on a self-attention mechanism known as transformer. The widely used wav2vec 2.0 is a self-supervised transformer model pre-trained on large amounts of unlabeled speech and then fine-tuned for a specific task. The data used for training and fine-tuning, along with the size of the transformer model, play a crucial role in both of these training steps. The most commonly used wav2vec 2.0 models are trained on relatively “clean” data from sources such as the LibriSpeech dataset, but we can expect there to be a benefit in using more realistic data gathered from a variety of acoustic conditions. However, it is not entirely clear how big the difference would be. Investigating this is the main goal of our article. To this end, we utilize wav2vec 2.0 models in three fundamental speech processing tasks: speaker change detection, voice activity detection, and overlapped speech detection, and test them on four real conversation datasets. We compare four wav2vec 2.0 models with different sizes and different data used for pre-training, and we fine-tune them either on in-domain data from the same dataset or on artificial training data created from the LibriSpeech corpus. Our results suggest that richer data that are more similar to the task domain bring better performance than a larger model.
- 16 September 2022
- Published by Springer Nature
- p. 377-388
Detection of Prosodic Boundaries in Speech Using Wav2Vec 2.0
- 16 September 2022
- Published by Springer Nature
- p. 377-388
Detection of Prosodic Boundaries in Speech Using Wav2Vec 2.0
- 22 September 2021
- Published by Springer Nature
- p. 807-817
Applying EEND Diarization to Telephone Recordings from a Call Center
- 22 September 2021
- Published by Springer Nature
- p. 807-817
Applying EEND Diarization to Telephone Recordings from a Call Center
- 15 September 2019
- Published by International Speech Communication Association
UWB-NTIS Speaker Diarization System for the DIHARD II 2019 Challenge
- 15 September 2019
- Published by International Speech Communication Association
UWB-NTIS Speaker Diarization System for the DIHARD II 2019 Challenge