Single channel speech denoising by DDPG reinforcement learning agent

<p dir="ltr">Speech denoising (SD) covers the algorithms that suppress the background noise from the contaminated speech and improve its clarity. In this paper, a novel SD algorithm is presented based on the deep deterministic policy gradient (DDPG) agent; an off-policy reinforcement...

وصف كامل

محفوظ في:
التفاصيل البيبلوغرافية
المؤلف الرئيسي: Sania Gul (18272227) (author)
مؤلفون آخرون: Muhammad Salman Khan (7202543) (author)
منشور في: 2025
الموضوعات:
الوسوم: إضافة وسم
لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!
_version_ 1864513531482734592
author Sania Gul (18272227)
author2 Muhammad Salman Khan (7202543)
author2_role author
author_facet Sania Gul (18272227)
Muhammad Salman Khan (7202543)
author_role author
dc.creator.none.fl_str_mv Sania Gul (18272227)
Muhammad Salman Khan (7202543)
dc.date.none.fl_str_mv 2025-07-19T15:00:00Z
dc.identifier.none.fl_str_mv 10.1016/j.apacoust.2025.110954
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/Single_channel_speech_denoising_by_DDPG_reinforcement_learning_agent/30971164
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Information and computing sciences
Artificial intelligence
Data management and data science
Machine learning
Speech denoising
Reinforcement learning
DDPG agentTD3 agent
Reward function
Time–frequency masking
dc.title.none.fl_str_mv Single channel speech denoising by DDPG reinforcement learning agent
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p dir="ltr">Speech denoising (SD) covers the algorithms that suppress the background noise from the contaminated speech and improve its clarity. In this paper, a novel SD algorithm is presented based on the deep deterministic policy gradient (DDPG) agent; an off-policy reinforcement learning (RL) agent with a continuous action space. The noisy speech is first converted from the time domain to the time–frequency (TF) domain by taking its short-time Fourier transform (STFT), and then two separate DDPG agents are trained on the magnitude and phase components of the STFT. The reward function used for training these agents is the relative perceptual quality score of speech. After training, the DDPG agents generate the magnitude and phase soft masks, when noisy speech is given as input to them. These masks are then applied to the complex STFT matrix of the noisy speech to obtain the denoised speech. For matched testing data, the proposed system offers an improvement of 1.55 points in the perceptual evaluation of speech quality (PESQ) over the unprocessed speech, the highest among the other recent state-of-the-art models used for comparison in this paper. It achieves this performance by utilizing data that is 7 times smaller than that required by other models. Also, its learnable parameters are the lowest among all models, almost 12 times less than the next most compact model, based on another continuous RL agent (policy gradient (PG)) for estimating its convolutional kernels. When cascaded with a coloured spectrogram-based SD model, the proposed model further improves the PESQ by 0.07, CSIG by 1.23, and COVL by 1.4 points; the metrics estimating respectively the perceived quality of speech, its composite distortion, and its composite overall quality, surpassing all other baseline systems compared here. In the cascaded configuration, our proposed model offers the highest gain in PESQ (by 0.78 points) at 50 times fewer episodes, compared to the already trained speech enhancement and recognition models utilizing discrete agents for their performance improvement.</p><h2 dir="ltr">Other Information</h2><p dir="ltr">Published in: Applied Acoustics<br>License: <a href="http://creativecommons.org/licenses/by/4.0/" target="_blank">http://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1016/j.apacoust.2025.110954" target="_blank">https://dx.doi.org/10.1016/j.apacoust.2025.110954</a></p>
eu_rights_str_mv openAccess
id Manara2_1f7409d3a9ae9ced80369443e1f35ecf
identifier_str_mv 10.1016/j.apacoust.2025.110954
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/30971164
publishDate 2025
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling Single channel speech denoising by DDPG reinforcement learning agentSania Gul (18272227)Muhammad Salman Khan (7202543)Information and computing sciencesArtificial intelligenceData management and data scienceMachine learningSpeech denoisingReinforcement learningDDPG agentTD3 agentReward functionTime–frequency masking<p dir="ltr">Speech denoising (SD) covers the algorithms that suppress the background noise from the contaminated speech and improve its clarity. In this paper, a novel SD algorithm is presented based on the deep deterministic policy gradient (DDPG) agent; an off-policy reinforcement learning (RL) agent with a continuous action space. The noisy speech is first converted from the time domain to the time–frequency (TF) domain by taking its short-time Fourier transform (STFT), and then two separate DDPG agents are trained on the magnitude and phase components of the STFT. The reward function used for training these agents is the relative perceptual quality score of speech. After training, the DDPG agents generate the magnitude and phase soft masks, when noisy speech is given as input to them. These masks are then applied to the complex STFT matrix of the noisy speech to obtain the denoised speech. For matched testing data, the proposed system offers an improvement of 1.55 points in the perceptual evaluation of speech quality (PESQ) over the unprocessed speech, the highest among the other recent state-of-the-art models used for comparison in this paper. It achieves this performance by utilizing data that is 7 times smaller than that required by other models. Also, its learnable parameters are the lowest among all models, almost 12 times less than the next most compact model, based on another continuous RL agent (policy gradient (PG)) for estimating its convolutional kernels. When cascaded with a coloured spectrogram-based SD model, the proposed model further improves the PESQ by 0.07, CSIG by 1.23, and COVL by 1.4 points; the metrics estimating respectively the perceived quality of speech, its composite distortion, and its composite overall quality, surpassing all other baseline systems compared here. In the cascaded configuration, our proposed model offers the highest gain in PESQ (by 0.78 points) at 50 times fewer episodes, compared to the already trained speech enhancement and recognition models utilizing discrete agents for their performance improvement.</p><h2 dir="ltr">Other Information</h2><p dir="ltr">Published in: Applied Acoustics<br>License: <a href="http://creativecommons.org/licenses/by/4.0/" target="_blank">http://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1016/j.apacoust.2025.110954" target="_blank">https://dx.doi.org/10.1016/j.apacoust.2025.110954</a></p>2025-07-19T15:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1016/j.apacoust.2025.110954https://figshare.com/articles/journal_contribution/Single_channel_speech_denoising_by_DDPG_reinforcement_learning_agent/30971164CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/309711642025-07-19T15:00:00Z
spellingShingle Single channel speech denoising by DDPG reinforcement learning agent
Sania Gul (18272227)
Information and computing sciences
Artificial intelligence
Data management and data science
Machine learning
Speech denoising
Reinforcement learning
DDPG agentTD3 agent
Reward function
Time–frequency masking
status_str publishedVersion
title Single channel speech denoising by DDPG reinforcement learning agent
title_full Single channel speech denoising by DDPG reinforcement learning agent
title_fullStr Single channel speech denoising by DDPG reinforcement learning agent
title_full_unstemmed Single channel speech denoising by DDPG reinforcement learning agent
title_short Single channel speech denoising by DDPG reinforcement learning agent
title_sort Single channel speech denoising by DDPG reinforcement learning agent
topic Information and computing sciences
Artificial intelligence
Data management and data science
Machine learning
Speech denoising
Reinforcement learning
DDPG agentTD3 agent
Reward function
Time–frequency masking