MULTI-INSTRUMENT DETECTION IN POLYPHONIC MUSIC USING GAUSSIAN MIXTURE BASED FACTORIAL HMM

We formulate the problem of detecting the constituent instruments in a polyphonic music piece as a joint decoding problem. From monophonic data, parametric Gaussian Mixture Hidden Markov Models (GM-HMM) are obtained for each instrument. We propose a method to use the above models in a factorial framework, termed as Factorial GM-HMM (F-GM-HMM). The states are jointly inferred to explain the evolution of each instrument in the mixture observation sequence. The dependencies are decoupled using variational inference technique. We show that the joint time evolution of all instruments' states can be captured using F-GM-HMM. We compare performance of proposed method with that of Student's-t mixture model (tMM) and GM-HMM in an existing latent variable framework. Experiments on two to five polyphony with 8 instrument models trained on the RWC dataset, tested on RWC and TRIOS datasets show that F-GM-HMM gives an advantage over the other considered models in segments containing co-occurring instruments.

Music-Theoretic Estimation of Chords and Keys from Audio

This paper proposes a new method for local key and chord estimation from audio signals. This method relies primarily on principles from music theory, and does not require any training on a corpus of labelled audio files. A harmonic content of the musical piece is first extracted by computing a set of chroma vectors. A set of chord/key pairs is selected for every frame by correlation with fixed chord and key templates. An acyclic harmonic graph is constructed with these pairs as vertices, using a musical distance to weigh its edges. Finally, the sequences of chords and keys are obtained by finding the best path in the graph using dynamic programming. The proposed method allows a mutual chord and key estimation. It is evaluated on a corpus composed of Beatles songs for both the local key estimation and chord recognition tasks, as well as a larger corpus composed of songs taken from the Billboard dataset.

Estimations, from tagging experiments, of mortality rates and other parameters respecting yellowfin and skipjack tuna

ENGLISH: In this paper, a method of analysis described by Gulland (1963) has been used to estimate the fishing mortality rates of tagged yellowfin and skipjack tuna for specific areas and years. Fishing mortality rates obtained for tagged tunas will also represent those for the entire population from which the tagged fishes were drawn, provided the assumptions used and corrections made for these analyses are valid. Total mortality rates of tagged fishes have also been computed. These are not assumed to be directly equivalent to the total mortality rates of the untagged populations,since tagged fishes are subject to additional types of attrition. These additional sources of mortality are also examined in this study. SPANISH: En el presente trabajo se ha usado un método de análisis descrito por Gulland (1963), para estimar las tasas de mortalidad de pesca de los atunes aleta amarilla y barrilete marcados en áreas y años específicos. Las tasas de mortalidad de pesca obtenidas en atunes marcados representarán también las de toda la población, de la cual fueron extraídos, previendo que las suposiciones usadas y las correcciones hechas para estos análisis sean válidas. Las tasas de mortalidad total de los peces marcados también han sido computadas. No se supone que éstas sean directamente equivalentes a las tasas de mortalidad total de las poblaciones no marcadas, ya que los peces marcados están sujetos también a otros tipos de pérdida. Estas otras causas de mortalidad son examinadas también en el presente estudio.

Report on the pinniped and sea otter tagging workshop 18-19 January 1979 Seattle, Washington

The Marine Mammal Tagging Office has been created by consensus of the agencies responsible for marine mammal management and the scientific community dealing with marine mammal tagging and marking. The purpose of ths office is to facilitate the dissemination of information with regard to tagging, marking, tags, and marks; to determine the need for new and better materials for tags; and to stimulate research, development, and testing programs. The American Institute of Biological Sciences was requested to coordinate a workshop to determine the status of pinniped tagging both nationally and internationally. Approximately 30 scientists were invited to participate in the workshop which was held on 18-19 January 1979 at the Sand Point Laboratory of the National Marine Fisheries Service in Seattle, Washington. Topics included ranged from specific tagging programs to general considerations and similar problems encountered by researchers. Participants also participated in one of three working groups -- Sea Otters, Phocids, and Otariids --to address pertinent issues. These break-out sessions resulted in the general recommendations and specific considerations sections of this report. Abstract authors include: Alton Y. Roppel; Ken Pitcher; Burney J. Le Boeuf; Wybrand Hoek; Robert M. Warneke; Don B. Siniff; Doug P. DeMaster; Daniel J. Miller; Ian Stirling; Roger L. Gentry; Lanny H. Cornell; James E. Antrim; Edward D. Asper; Mark Keyes; R. Keith Farrell; Donald G. Calkins; Bob DeLong; T. A. Gornall; Tom Otten; and, Ancel M. Johnson (PDF contains 54 pages)

Knowledge Exchange study: How Research Tools are of Value to Research: Use Cases and Recommendations

Research tools that are freely available and accessible via the Internet cover an emergent field in the worldwide research infrastructure. Clearly, research tools have increasing value for researchers in their research activities. Knowledge Exchange recently commissioned a project to explore use case studies to show research tools’ potential and relevance for the present research landscape. Makers of successful research tools have been asked questions such as: How are these research tools developed? What are their possibilities? How many researchers use them? What does this new phenomenon mean for the research infrastructure? Additional to the Use Cases, the authors offer observations and recommendations to contribute to effective development of a research infrastructure that can optimally benefit from research tools. the Use Cases are: •Averroes Goes Digital: Transformation, Translation, Transmission and Edition •BRIDGE: Tools for Media Studies Researchers •Multiple Researchers, Single Platform: A Virtual Tool for the 21st Century •The Fabric of Life •Games with A Purpose: How Games Are Turning Image Tagging into Child’s Play •Elmer: Modelling a Future •Molecular Modelling With SOMA2 •An Online Renaissance for Music: Making Early Modern Music Readable •Radio Recordings for Research: How A Million Hours of Danish Broadcasts Were Made Accessible •Salt Rot: A Central Space for Essential Research •Cosmos: Opening Up Social Media for Social Science A brief analysis by the authors can be found: •Some Observations Based on the Case Studies of Research Tools

Blood lactate in yellowfin tuna, Neothunnus macropterus, and skipjack, Katsuwonus pelamis, following capture and tagging

ENGLISH: The staff of the Inter-American Tropical Tuna Commission for several years has been investigating the life history, population structure, behavior and ecology of the yellowfin tuna, Neothunnus macropterus, and the skipjack, Katsuwonus pelamis, in the Eastern Tropical Pacific Ocean. The tagging and subsequent recovery of these tropical tunas, to provide information on population structure, migrations, mortality rates and growth rates, are important aspects of these investigations. Broadhead (1959) and Schaefer, Chatwin and Broadhead (1961) emphasize the many difficulties involved in tagging these extremely active yet delicate fish and give considerable evidence to suggest that tagging mortality is high, perhaps as great as 60 to 80 per cent. The latter authors suggest that the rather high mortality at tagging is related to the effects of hyperactivity brought about by the tagging operation. SPANISH: El personal de la Comisión Interamericana del Atún Tropical ha estado investigando durante varios años la historia natural, la estructura de la población, los hábitos y la ecología del atún aleta amarilla, Neothunnus macropterus, y del barrilete, Katsuwonus pelamis, en el Océano Pacífico Oriental Tropical. La marcación y el subsiguiente recobro de estos atunes tropicales, lo que da información sobre la estructura de la población, los movimientos migratorios y las tasas de crecimiento y de mortalidad, son importantes aspectos de estas investigaciones. Broadhead (1959) y Schaefer, Chatwin y Broadhead (1961) destacan las muchas dificultades que hay para marcar estos peces activos en extremo pero delicados, y proporcionan considerable evidencia que sugiere que la mortalidad por la marcación es bastante alta, siendo quizás de 60 a 80 por ciento. Los autores citados sugieren que esta elevada mortalidad por la marcación está relacionada con los efectos de la hiperactividad producida por la operación de marcación.

Tagging and recovery of tropical tunas 1955-1959

ENGLISH: Yellowfin and skipjack tuna occur in commercial quantities in the Eastern Pacific Ocean from California to Chile. They are captured in the high seas at distances from the mainland up to several hundred miles (see Alverson, 1960). The Inter-American Tropical Tuna Commission has been engaged for several years in research on the biology, ecology, and population dynamics of the stocks of these species supporting the commercial fishery, in order to elucidate the effects of the fishery and of fishery independent factors on their abundance and behavior, to provide the scientific basis for rational management of the fishery. An important aspect of this research is the investigation of the migrations of these species in the Eastern Pacific, and the determination of whether each consists of but a single population or is composed of various sub-populations. One direct means of approaching these problems is the tagging, and subsequent recovery, of specimens in the region of the commercial fishery. This also provides direct information on growth rates, by comparison of sizes of specimens at tagging and upon later recovery, and can furnish the basis of estimating rates of mortality. These are two of the important elements of the vital statistics of the tuna populations. SPANISH: El atún aleta amarilla y el barrilete se encuentran en cantidades comerciales en el Océano Pacífico Oriental, desde California hasta Chile. Estos peces son capturados en alta mar a varios cientos de millas de distancia de tierra firme (ver Alverson, 1960). La Comisión Interamericana del Atún Tropical ha estado dedicada durante varios años a la investigación de la biología, ecología y dinámica de las poblaciones de los stocks de las indicadas especies que mantienen la pesquería comercial, a fin de elucidar los efectos de ésta y de los factores independientes de la explotación sobre la abundancia y hábitos de estos peces, para obtener una base científica que permita una administración racional de la pesquería. Un aspecto importante de esta investigación es el estudio de los movimientos migratorios de estas especies en el Pacífico Oriental, y la determinación de que si cada una constituye una sola población o está compuesta de varias subpoblaciones. Un medio directo de abordar estos problemas es el de la marcación, y subsecuente recuperación, de especímenes en la región de la pesquería comercial. Esto también proporciona una información directa sobre la tasa de crecimiento, por la comparación de los tamaños de los especímenes al ser marcados y recuperados más tarde y puede proveer la base para estimar las tasas de mortalidad. Estos son dos de los elementos importantes de las estadísticas vitales de las poblaciones de atún.

Muscle glycogen and blood lactate in yellowfin tuna, Thunnus albacares, and skipjack, Katsuwonus pelamis, following capture and tagging

ENGLISH: Tagging and the recovery of tagged yellowfin (Thunnus albacares) and skipjack (Katsuwonus pelamis) tunas are important aspects of the investigations conducted by the Inter-American Tropical Tuna Commission in the Eastern Tropical Pacific Ocean. The results of the tagging program provide information on population structures, migrations, mortality rates and growth rates of these two species. The present experimental program was undertaken to study the relationship between muscular fatigue and high tagging mortalities in yellowfin and skipjack. SPANISH: La marcación del atún aleta amarilla (Thunnus albacares) y del barrilete (Katsuwonus pelamis), y el recobro de estos atunes marcados, son aspectos importantes de la investigación que efectúa la Comisión Interamericana del Atún Tropical en el Océano Pacífico Oriental Tropical. Los resultados del programa de marcación proporcionan información sobre la estructura de las poblaciones, migraciones, tasas de mortalidad y tasas de crecimiento de estas dos especies. El programa experimental presente fue emprendido para estudiar la relación entre la fatiga muscular y la alta mortalidad causada por la marcación en el atún aleta amarilla y el barrilete. (PDF contains 52 pages.)

Population dynamics of the anchoveta, Cetengraulis mysticetus, in the Gulf of Panama, as determined by tagging experiments

ENGLISH: The anchoveta is the major constituent of the important bait and reduction fisheries of the Gulf of Panama. It is a short-lived species, the great majority of the catch consisting of fish in their first year of life. Fish for reduction are caught chiefly in the Isla Verde area, between Punta Chame and the entrance of the Panama Canal. In 1960 and 1961 anchovetas were tagged with metal internal tags and released in the major areas of occurrence of this species. The tags were recovered from the meal in the reduction plants with magnets. From the 53,380 fish tagged in 1960, 745 tags were returned during the 1960 season, 246 during the 1961 season, and 8 during the 1962 season. From the 113,202 tagged in 1961, 373 tags were returned during the 1961 season and 48 during the 1962 season. Complete catch statistics are available, and length-frequency and length-weight data were used to convert these from pounds to numbers of fish of each year class. The annual survival rate for the fish of the 1959 year class in the Isla Verde area was estimated to be 0.086 by the Chapman-Robson method, 0.102 by the year-class method, and 0.088 by the Jackson positive method. The first method is considered to give the best estimate. Six estimates of the population of fish of the 1959 year class in the Isla Verde area were obtained from the sample tag ratios of six experiments conducted in that area in 1960. The estimates differed due to the temporal decrease in the population, but the downward trend corresponded fairly well to what was expected from the total annual mortality rate. It was estimated that the population of 1959-year class fish was about 818 million on March 8, 1960, and about 70 million on March 8, 1961. As the population of anchovetas decreases during the season the effort increases sufficiently that the catch remains roughly constant. This is described as the "constant absolute catch" type fishery. Of the original population of fish in the Isla Verde area at the beginning of the 1960 season, about 11 per cent were caught and 81 per cent died of natural causes. Evaluation of growth and mortality data demonstrated that beginning the fishery for the youngest age group later than March 8 (the date it began in 1960) would reduce the yield per recruit, while increasing the fishing effort would greatly increase it. Further, it is believed unlikely that increases in the catch in the Isla Verde area alone would noticeably decrease the number of recruits to that area. Therefore there is no foreseeable need for regulation of the fishery. SPANISH: El principal constituyente de la importante pesquería para carnada y para reducción en el Golfo de Panamá es la anchoveta. Es una especie de vida corta cuya pesca, en su mayor parte, está constituida por peces que se encuentran en su primer año de vida. Para la industria de reducción los peces son capturados principalmente en el área de Isla Verde, entre Punta Chame y la entrada del Canal de Panamá. En 1960 y 1961 las anchovetas fueron marcadas con marcas metálicas internas y liberadas en las áreas más importantes en que se encuentra esta especie. Las marcas fueron recobradas de la harina en las plantas de reducción por medio de magnetos. De los 53,380 peces marcados en 1960, fueron devueltas 745 marcas durante la temporada pesquera de 1960, 246 durante la de 1961, y 8 durante la de 1962. De los 113,202 marcados en 1961, 373 marcas fueron devueltas durante la temporada pesquera de 1961 y 48 durante la de 1962. Se dispone de estadísticas completas de captura, y los datos de frecuencia-longitud y de longitud-peso fueron usados para convertir éstos de libras a números de peces de cada clase anual. La tasa anual de supervivencia correspondiente a la clase anual de 1959 en el área de Isla Verde estimó en 0.086 por medio del método Chapman-Robson; en 0.102 por método de la clase anual; y en 0.088 por el método positivo de Jackson. Se considera que el primer método dé la mejor estimación. Seis estimaciones de la población de peces de la clase anual 1959 en el área de Isla Verde fueron obtenidas según la proporción de marcas halladas en las muestras correspondientes a seis experimentos efectuados en aquella área en 1960. Las estimaciones variaron debido a la disminución temporal de la población, pero esta tendencia descendente correspondió bastante bien a lo que se esperaba según la tasa total de mortalidad anual. Se estimó que la población de peces de la clase anual de 1959 era de unos 818 millones el 8 de marzo de 1960, y aproximadamente de unos 70 millones el 8 de marzo de 1961. Conforme a que la población de anchovetas disminuye durante la temporada pesquera, el esfuerzo aumenta lo suficientemente como para que la pesca se mantenga más o menos constante. Este es el tipo de pesquería descrito como de "captura absoluta constante". De la población original de peces en el área de Isla Verde al comienzo de la temporada pesquera de 1960, cerca del 11 por ciento fue capturada y el 81 por ciento murió por causas naturales. La evaluación de los datos del crecimiento mortalidad demostraron que al comenzar la pesquería a explotar grupo de edad más joven en una fecha posterior al 8 de marzo (la fecha en que comenzó en 1960) se reduciría el rendimiento por recluta, mientras que al aumentar el esfuerzo de pesca lo aumentaría considerablemente. Más aún, se cree improbable que el aumento en la pesca en el área de Isla Verde de por sí disminuyera perceptiblemente el número de reclutas en esa área. En consecuencia no se prevé la necesidad de una reglamentación de la pesquería. (PDF contains 172 pages.)

Estimates of the rates of mortality of yellowfin tuna in the Eastern Pacific Ocean derived from tagging experiments

ENGLISH: Tag release and return data for the Baja California and Gulf of Guayaquil areas were selected for this study because substantial numbers of returns resulted from these releases and because the effects of emigration are small in these areas. The returns of tags per unit of fishing effort for several experiments in each area were used to estimate the coefficients of total mortality and shedding. The coefficient of annual natural mortality was estimated to be less than 2.0, which is in agreement with a previous estimate of 0.8, but does not improve upon it. The estimates for the average coefficients of catchability are 2.02 X 10-3 for the Baja California area and 0.67 X 10-3 for the Gulf of Guayaquil area. SPANISH: Se seleccionaron para este estudio algunos da tos de liberación y retorno de marcas en las áreas de Baja California y el Golfo de Guayaquil debido a que cantidades substanciales de retornos resultaron de estas liberaciones y porque los efectos de migración son pequeños en estas áreas. Los retornos de marcas por unidad de esfuerzo de pesca de varios experimentos en cada área fueron empleados para estimar los coeficientes de mortalidad total y desprendimiento. Se estimó que el coeficiente de mortalidad natural anual fue inferior a 2.0, lo que está de acuerdo con una estimación anterior de 0.8, pero no la mejora. Las estimaciones de los coeficientes promedios de capturabilidad son 2.02 X 10-3 en el área de Baja California y 0.67 X 10-3 en el área del Golfo de Guayaquil. (PDF contains 58 pages.)

Materials and methods for tagging purse seine and baitboat caught tunas

ENGLISH: The percentages of return of purse seine-tagged yellowfin and baitboat- tagged yellowfin and skipjack subjected to various combinations of treatments were tested statistically to determine the effects of the treatments. For yellowfin the precentages of return differed among sets or dates of release, and were higher for double-tagged fish, larger fish, and fish released soonest during each set. For skipjack the percentages of return differed among dates of release, and were higher for single-tagged fish and fish tagged in plastic-covered cradles. SPANISH: Los porcentajes de retorno de aleta amarilla marcado en barcos cerqueros y los de aleta amarilla y barrilete marcados en los de carnada sometidos a varias combinaciones de tratamientos fueron probados estadísticamente para determinar los efectos de los tratamientos. Con respecto al aleta amarilla los porcentajes de retorno fueron diferentes entre los lances o fechas de liberación, y fueron más altos para los peces con doble marca, peces de más talla y peces liberados lo más pronto durante cada lance. Los porcentajes de retorno del barrilete fueron diferentes entre las fechas de liberación, y fueron más altos para peces con una sola marca y peces marcados en cunas cubiertas con plástico. (PDF contains 41 pages.)