TCD-VoIP, a research database of degraded speech for assessing quality in VoIP applications

Harte, Naomi; Gillen, Eoin; Hines, Andrew

doi:10.1109/qomex.2015.7148100

Cited by 27 publications

(44 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We used the TCD-VoIP corpora [12] to evaluate our model. This database contains subjective quality ratings in the presence of different degradations that can occur in VoIP applications, which are not limited to narrowband data.…”

Section: Subjective Listening Datamentioning

confidence: 99%

See 1 more Smart Citation

Prediction of Perceived Speech Quality Using Deep Machine Listening

2018

View full text Add to dashboard Cite

Subjective ratings of speech quality (SQ) are essential for evaluating algorithms for speech transmission and enhancement. In this paper we explore a non-intrusive model for SQ prediction based on the output of a deep neural net (DNN) from a regular automatic speech recognizer. The degradation of phoneme probabilities obtained from the net is quantified with the mean temporal distance proposed earlier for multi-stream ASR. The SQ predicted with this method is compared with average subject ratings from the TCD-VoIP speech quality database that covers several effects of SQ degradation that can occur in VoIP applications such as clipping, packet loss, echo effects, background noise, and competing speakers. Our approach is tailored to speech and therefore not applicable when quality is degraded by a competing speaker, which is reflected by an insignificant correlation between model output and subjective SQ. In all other conditions mentioned above, the model reaches an average correlation of r = 0.87, which is higher than the correlation achieved with the baseline ITU-T P.563 (r = 0.71) and the American National Standard ANIQUE+ (r = 0.75). Since the most robust ASR system is not necessarily the best model to predict SQ, we investigate the effect of the amount of training data on quality prediction.

show abstract

Section: Subjective Listening Datamentioning

confidence: 99%

“…Our model is evaluated using the TCD-VoIP database that contains subjective ratings for signals distorted by VoIP artifacts [12]. We analyze the correlation between the mean objective scores (MOS) [13] and the model output, and compare the results to two baseline measures (ANIQUE+ and ITU P.563).…”

Section: Introductionmentioning

confidence: 99%

Prediction of Perceived Speech Quality Using Deep Machine Listening

2018

View full text Add to dashboard Cite

show abstract

“…The TCD-VoIP dataset contemplates a number of degradations that commonly occur in a voice over IP transmission environment. Such degradations are referred to as platformindependent, that is, degradations that occur independently of the codec, hardware, or network in use [8]. The dataset contains five types of degradations: 1) background noise, 2) competing speakers, 3) echo effects, 4) amplitude clipping, and 5) choppy speech.…”

Section: The Tcd-voip Datasetmentioning

confidence: 99%

“…Quality scores were gathered for a set of audio-visual sequences with distortions only in the audio component. The TCD-VoIP dataset [8] served as a reference to produce a new audio-visual dataset with onlyaudio distortions: the Im-AV-Exp2. The experiment had the goal of recreating some of the streaming audio degradations from the TCD-VoIP dataset on an audio-visual scenario and analyzing the effect of such degradations on the perceived audio-visual quality.…”

Section: Introductionmentioning

confidence: 99%

Perceived quality of audio-visual stimuli containing streaming audio degradations

Martinez

Farias

Hines

2018

2018 26th European Signal Processing Conference (EUSIPCO)

Self Cite

View full text Add to dashboard Cite

Multimedia services play an important role in modern human communication. Understanding the impact of multisensory input (audio and video) on perceived quality is important for optimizing the delivery of these services. This work explores the impact of audio degradations on audiovisual quality. With this goal, we present a new dataset that contains audiovisual sequences with distortions only in the audio component (Im-AV-Exp2). The degradations in this new dataset correspond to commonly encountered streaming degradations, matching those found in the audio-only TCD-VoIP dataset. Using the Immersive Methodology, we perform a subjective experiment with the Im-AV-Exp2 dataset. We analyze the experimental data and compared the quality scores of the Im-AV-Exp2 and TCD-VoIP datasets. Results show that the video component act as a masking factor for certain classes of audio degradations (e.g. echo), showing that there is an interaction of video and audio quality that may depend on content.

show abstract

“…Esses problemas originam degradações no sinal transmitido, que para o caso de uma chamada VoIP, os principais são: a taxa de perda de pacotes (ou PLR, do inglês, Packet Loss Rate), atrasos e variações de atrasos. Existem diversos estudos [1][2][3] O modelo de Gilbert-Elliot [4], baseado na cadeia de Markov de dois estados, é utilizado para determinar diversos padrões de perda de pacotes, ou seja, perdas concentradas em curtos segmentos de tempo, conhecidas como rajadas, e as perdas distribuídas em segmento de maior duração, ambas contendo o mesmo número de pacotes perdidos no segmento total de voz estudado. Para a determinação do índice MOS, métodos objetivos e subjetivos de avaliação de qualidade de voz são utilizados.…”

Section: Iunclassified

Estudo da Distribuição Temporal das Perdas de Pacotes e o Seu Impacto na Qualidade de Voz

Ieee¹,

Pereira²,

Rodríguez³

2016

Anais Do XXXIV Simpósio Brasileiro De Telecomunicações

View full text Add to dashboard Cite

Resumo-O serviço de comunicações de voz sobre IP (VoIP) sofre diferentes tipos de degradações na transmissão, como perda de pacotes, atrasos e variações de atrasos. Diversos estudos alegam que a perda de pacotes é um dos fatores mais degradantes; porém, o impacto da distribuição e a localização temporal dessas perdas em um segmento de fala não foram determinados. Neste contexto, este artigo apresenta resultados que demostram a importância de conhecer a distribuição das perdas de pacotes para determinar com maior exatidão um índice de qualidade MOS.

show abstract

TCD-VoIP, a research database of degraded speech for assessing quality in VoIP applications

Cited by 27 publications

References 14 publications

Prediction of Perceived Speech Quality Using Deep Machine Listening

Prediction of Perceived Speech Quality Using Deep Machine Listening

Perceived quality of audio-visual stimuli containing streaming audio degradations

Estudo da Distribuição Temporal das Perdas de Pacotes e o Seu Impacto na Qualidade de Voz

Contact Info

Product

Resources

About