Speaker linking using complete-linkage clustering

Speaker diarization determines instances of the same speaker within a recording. Extending this task to a collection of recordings for linking together segments spoken by a unique speaker requires speaker linking. In this paper we propose a speaker linking system using linkage clustering and state-of-the-art speaker recognition techniques. We evaluate our approach against two baseline linking systems using agglomerative cluster merging (AC) and agglomerative clustering with model retraining (ACR). We demonstrate that our linking method, using complete-linkage clustering, provides a relative improvement of 20% and 29% in attribution error rate (AER), over the AC and ACR systems, respectively.

Speaker attribution of multiple telephone conversations using a complete-linkage clustering approach

In this paper we propose and evaluate a speaker attribution system using a complete-linkage clustering method. Speaker attribution refers to the annotation of a collection of spoken audio based on speaker identities. This can be achieved using diarization and speaker linking. The main challenge associated with attribution is achieving computational efficiency when dealing with large audio archives. Traditional agglomerative clustering methods with model merging and retraining are not feasible for this purpose. This has motivated the use of linkage clustering methods without retraining. We first propose a diarization system using complete-linkage clustering and show that it outperforms traditional agglomerative and single-linkage clustering based diarization systems with a relative improvement of 40% and 68%, respectively. We then propose a complete-linkage speaker linking system to achieve attribution and demonstrate a 26% relative improvement in attribution error rate (AER) over the single-linkage speaker linking approach.

Complete-linkage clustering for voice activity detection in audio and visual speech

We propose a novel technique for conducting robust voice activity detection (VAD) in high-noise recordings. We use Gaussian mixture modeling (GMM) to train two generic models; speech and non-speech. We then score smaller segments of a given (unseen) recording against each of these GMMs to obtain two respective likelihood scores for each segment. These scores are used to compute a dissimilarity measure between pairs of segments and to carry out complete-linkage clustering of the segments into speech and non-speech clusters. We compare the accuracy of our method against state-of-the-art and standardised VAD techniques to demonstrate an absolute improvement of 15% in half-total error rate (HTER) over the best performing baseline system and across the QUT-NOISE-TIMIT database. We then apply our approach to the Audio-Visual Database of American English (AVDBAE) to demonstrate the performance of our algorithm in using visual, audio-visual or a proposed fusion of these features.

Robust automatic speaker linking and attribution

This research makes a major contribution which enables efficient searching and indexing of large archives of spoken audio based on speaker identity. It introduces a novel technique dubbed as “speaker attribution” which is the task of automatically determining ‘who spoke when?’ in recordings and then automatically linking the unique speaker identities within each recording across multiple recordings. The outcome of the research will also have significant impact in improving the performance of automatic speech recognition systems through the extracted speaker identities.

Speaker attribution of Australian broadcast news data

Speaker attribution is the task of annotating a spoken audio archive based on speaker identities. This can be achieved using speaker diarization and speaker linking. In our previous work, we proposed an efficient attribution system, using complete-linkage clustering, for conducting attribution of large sets of two-speaker telephone data. In this paper, we build on our proposed approach to achieve a robust system, applicable to multiple recording domains. To do this, we first extend the diarization module of our system to accommodate multi-speaker (>2) recordings. We achieve this through using a robust cross-likelihood ratio (CLR) threshold stopping criterion for clustering, as opposed to the original stopping criterion of two speakers used for telephone data. We evaluate this baseline diarization module across a dataset of Australian broadcast news recordings, showing a significant lack of diarization accuracy without previous knowledge of the true number of speakers within a recording. We thus propose applying an additional pass of complete-linkage clustering to the diarization module, demonstrating an absolute improvement of 20% in diarization error rate (DER). We then evaluate our proposed multi-domain attribution system across the broadcast news data, demonstrating achievable attribution error rates (AER) as low as 17%.

A speaker rediarization scheme for improving diarization in large two-speaker telephone datasets

In this paper we propose a novel scheme for carrying out speaker diarization in an iterative manner. We aim to show that the information obtained through the first pass of speaker diarization can be reused to refine and improve the original diarization results. We call this technique speaker rediarization and demonstrate the practical application of our rediarization algorithm using a large archive of two-speaker telephone conversation recordings. We use the NIST 2008 SRE summed telephone corpora for evaluating our speaker rediarization system. This corpus contains recurring speaker identities across independent recording sessions that need to be linked across the entire corpus. We show that our speaker rediarization scheme can take advantage of inter-session speaker information, linked in the initial diarization pass, to achieve a 30% relative improvement over the original diarization error rate (DER) after only two iterations of rediarization.

A cluster-voting approach for speaker diarization and linking of Australian broadcast news recordings

We present a clustering-only approach to the problem of speaker diarization to eliminate the need for the commonly employed and computationally expensive Viterbi segmentation and realignment stage. We use multiple linear segmentations of a recording and carry out complete-linkage clustering within each segmentation scenario to obtain a set of clustering decisions for each case. We then collect all clustering decisions, across all cases, to compute a pairwise vote between the segments and conduct complete-linkage clustering to cluster them at a resolution equal to the minimum segment length used in the linear segmentations. We use our proposed cluster-voting approach to carry out speaker diarization and linking across the SAIVT-BNEWS corpus of Australian broadcast news data. We compare our technique to an equivalent baseline system with Viterbi realignment and show that our approach can outperform the baseline technique with respect to the diarization error rate (DER) and attribution error rate (AER).

Extending the task of diarization to speaker attribution

In this paper we extend the concept of speaker annotation within a single-recording, or speaker diarization, to a collection wide approach we call speaker attribution. Accordingly, speaker attribution is the task of clustering expectantly homogenous intersession clusters obtained using diarization according to common cross-recording identities. The result of attribution is a collection of spoken audio across multiple recordings attributed to speaker identities. In this paper, an attribution system is proposed using mean-only MAP adaptation of a combined-gender UBM to model clusters from a perfect diarization system, as well as a JFA-based system with session variability compensation. The normalized cross-likelihood ratio is calculated for each pair of clusters to construct an attribution matrix and the complete linkage algorithm is employed to conduct clustering of the inter-session clusters. A matched cluster purity and coverage of 87.1% was obtained on the NIST 2008 SRE corpus.

Técnicas multivariadas na determinação da diversidade genética em gergelim usando marcadores RAPD

O objetivo deste trabalho foi comparar diferentes técnicas multivariadas na caracterização de 35 genótipos de gergelim mediante 769 marcadores RAPD. As distâncias genéticas foram obtidas pelo complemento aritmético do coeficiente de Jaccard e agrupadas pelos métodos hierárquicos do vizinho mais próximo, do vizinho mais distante, das médias aritméticas não ponderadas (UPGMA), do método de otimização de Tocher e análises de coordenadas principais. O agrupamento dos genótipos foi alterado em função dos diferentes métodos usados. Adotando-se a mesma distância genética (0,36) como valor de corte, diferenciaram-se quatro grupos no método do vizinho mais próximo, 13 para o vizinho mais distante, 11 no UPGMA e quatro no Tocher. Entre os métodos hierárquicos, o UPGMA apresentou o melhor ajuste das distâncias originais e estimadas (CCC = 0,89). As análises das coordenadas principais confirmaram a baixa diversidade existente entre os genótipos. A maior divergência ocorreu entre as cultivares Seridó 1 e Arawaca 4, e a menor, entre os genótipos VCR-101 e GP-3314. As três primeiras coordenadas principais contabilizaram 35,13% do total da variabilidade, e 18 autovalores foram necessários para explicar 81% da variação genética. Os métodos UPGMA, de otimização de Tocher, e as análises de coordenadas principais são complementares na formação dos grupos.

Tamanho de amostra para a estimativa das médias decendiais de radiação solar global no Estado do Rio Grande do Sul

Com o objetivo de verificar a variabilidade temporal e espacial do tamanho de amostra da radiação solar global média decendial, de 22 locais do Estado do Rio Grande do Sul, utilizaram-se séries de dados de radiação solar global do período de 1956 a 2003. Determinou-se o tamanho de amostra da radiação solar global média decendial em cada decêndio e local e agruparam-se os decêndios e os locais pelo método hierárquico 'vizinho mais distante'. Há variabilidade do tamanho de amostra (número de anos) para a estimativa da radiação solar global média decendial no Estado do Rio Grande do Sul no tempo e no espaço. Maior tamanho é necessário nos decêndios dos meses de junho, julho, agosto e setembro em relação aos outros meses. Para os locais e decêndios estudados, 30 anos de observações são suficientes para estimar a média (µ) de radiação solar global média decendial, para um erro de estimação igual a 12.3%, com coeficiente de confiança de 95%.

PHYLOGENETIC STUDIES OF SOME SPECIES OF THE GENUS COFFEA .1. NUMERICAL-ANALYSIS OF FLAVONOID COMPOUNDS

Flavonoid compounds were analyzed in ripe fruit pulp of ten species of Coffea, including two cultivars of C. arabica and two of C. canephora. Three coefficients of similarity: Simple-Matching, Jaccard and Ochiai and three different clustering methods, Single Linkage, Complete Linkage and Unweighted Pair Group, Using Arithmetic Averages (UPGMA), were used to analyze the data.Jaccard and Ochiai's coefficients of association showed a more coherent result, when compared with taxonomic and hybridization studies. Inclusion of Psilanthopsis kapakata in the genus Coffea, as C. kapakata, is justified by the similarity of this species with other studied species, and clusters clearly approximate the species C. arabica and C. eugenioides. The latter is one of the possible parents of the allotetraploid species C. arabica, C. congensis is the only species whose position remains ambiguous, probably due to the fact that the plants of this species that were introduced into the Campinas collections, were hybrids and not typical of C. congensis.

PHYLOGENETIC STUDIES OF SOME SPECIES OF THE GENUS COFFEA .2. NUMERICAL-ANALYSIS OF ISOENZYMATIC DATA

Thirteen species of Coffea were studied for five enzymes systems, including alpha and beta esterase, alkaline phosphatase, acid phosphatase, malate dehydrogenase and acid dehydrogenase. Three coefficients of similarity: Simple Matching, Jaccard and Ochiai and three different clustering methods: Single Linkage, Complete Linkage and Unweighted Pair Group, using Arithmetic Averages (UPGMA) were used to analyse the data.The phylogenetic relationships among the twelve diploid species and between them and the tetraploid species C. arabica showed that similarity among species of the same subsection is not always greater than among species of different subsections. In addition, although there are several similarity groups in common, established by isoenzymatic polymorphism, morphological characteristics, chemical data, crossability and geographic distribution, there is no common trend among the phylogenetic relationships as indicated by all these different evaluating procedures.

Avaliação da base genética de genótipos de soja resistentes ao nematóide de cisto (raça 3) por marcadores microssatélites

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Heuristics for minimizing the maximum within-clusters distance

The clustering problem consists in finding patterns in a data set in order to divide it into clusters with high within-cluster similarity. This paper presents the study of a problem, here called MMD problem, which aims at finding a clustering with a predefined number of clusters that minimizes the largest within-cluster distance (diameter) among all clusters. There are two main objectives in this paper: to propose heuristics for the MMD and to evaluate the suitability of the best proposed heuristic results according to the real classification of some data sets. Regarding the first objective, the results obtained in the experiments indicate a good performance of the best proposed heuristic that outperformed the Complete Linkage algorithm (the most used method from the literature for this problem). Nevertheless, regarding the suitability of the results according to the real classification of the data sets, the proposed heuristic achieved better quality results than C-Means algorithm, but worse than Complete Linkage.

Prescriptions of Chinese medicinal herbs in Switzerland: the example of suan zao ren (Ziziphi Spinosae Semen)

Introduction According to the Swiss Health Survey 2007, 1.7% of the adult population use traditional Chinese medicine (including Chinese herbal medicine, but excluding acupuncture). In contrast to conventional drugs, that contain single chemically defined substances, prescriptions of Chinese herbs are mixtures of up to 40 ingredients (parts of plants, fungi, animal substances and minerals). Originally they were taken in the form of decoctions, but nowadays granules are more popular. Medium daily dosages of granules range between 8 to 12g. In a recent work we identified the most commonly used Chinese herbs (all ingredients are referred to as herbs for reasons of simplicity) and classical formulas (mixtures). Here we present a short overview and the example of suan zao ren (Ziziphi Spinosae Semen), which is used in the treatment of insomnia and anxiety and contains saponins that have been shown to increase sleep in animal studies. Material and Methods A random sample of 1,053 prescriptions was drawn from the database of Lian Chinaherb AG, Switzerland, and analysed according to the most frequently used individual herbs and classical formulas. Cluster analysis (Jaccard similarity coefficient, complete linkage method) was applied to identify common combinations of herbs. Results The most frequently used herbs were dang gui (Angelicae Sinensis Radix), fu ling (Poria), bai shao (Paeoniae Radix Alba), and gan cao (Glycyrrhizae Radix et Rhizoma); the most frequently used classical formulas were gui pi tang (Restore the Spleen Decoction) and xiao yao san (Rambling Powder). The average number of herbs per prescription was 12.0, and the average daily dosage of granules was 8.7g. 74.3% of the prescriptions were for female, 24.8% for male patients. Suan zao ren was present in 14.2% of all prescriptions. These prescriptions contained on average 13.7 herbs, and the daily dosage of granules was 8.9g. Suan zao ren was more frequently prescribed by practitioners of non-Asian than of Asian origin but equally often for female and male patients. Cluster analysis grouped suan zao ren with yuan zhi (Polygalae Radix), bai zi ren (Platycladi Semen), sheng di huang (Rehmanniae Radix) and dan shen (Salviae Miltiorrhizae Radix et Rhizoma). Discussion Prescriptions including suan zao ren contained on average slightly more herbs than other prescriptions. This might be due to the fact that two of the three most popular classical formulas with suan zao ren are composed of 13 and 12 herbs with the possibility of adding more ingredients when necessary. Cluster analysis resulted in the clustering of suan zao ren with other herbs of the classical formula tian wang bu xin dan (Emperor of Heaven’s Special Pill to Tonify the Heart), indicating the use of suan zao ren for the treatment of insomnia and irritability. Unfortunately, the diagnoses of the patients were unavailable and thus correlations between use of suan zao ren and diseases could not be analysed.