Hash tables – which map “keys ” onto “values” – are an essential building block in modern software systems. We believe a similar functionality would be equally valuable to large distributed systems. In this paper, we introduce the concept of a Content-Addressable Network (CAN) as a distributed infrastructure that provides hash table-like functionality on Internet-like scales. The CAN is scalable, fault-tolerant and completely self-organizing, and we demonstrate its scalability, robustness and low-latency properties through simulation.

... describe music. They are commonly used to structure the increasing amounts of music available in digital form on the Web and are important for music information retrieval. Genre categorization for audio has traditionally been performed manually. A particular musical genre is characterized by statistical properties related to the instrumentation, rhythmic structure and form of its members. In this work, algorithms for the automatic genre categorization of audio signals are described. More specifically, we propose a set of features for representing texture and instrumentation. In addition a novel set of features for representing rhythmic structure and strength is proposed. The performance of those feature sets has been evaluated by training statistical pattern recognition classifiers using real world audio collections. Based on the automatic hierarchical genre classification two graphical user interfaces for browsing and interacting with large audio collections have been developed.

Abstract not found

Abstract. We report on experiments done in an attempt to improve the performance of a music similarity measure which we introduced earlier. The technique aims at comparing music titles on the basis of their global “timbre”, which has many applications in the field of Music Information Retrieval. Such measures of timbre similarity have seen a growing interest lately, and every contribution (including ours) is yet another instantiation of the same basic pattern recognition architecture, only with different algorithm variants and parameters. Most give encouraging results with a little effort, and imply that near-perfect results would just extrapolate by fine-tuning the algorithms ’ parameters. However, such systematic testing over large, interdependent parameter spaces is both difficult and costly, as it requires to work on a whole general meta-database architecture. This paper contributes in two ways to the current state of the art. We report on extensive tests over very many parameters and algorithmic variants, either already envisioned in the literature or not. This leads to an improvement over existing algorithms of about 15 % R-precision. But most importantly, we describe many variants that surprisingly do not lead to any substancial improvement. Moreover, our simulations suggest the existence of a “glass ceiling ” at R-precision about 65 % which cannot probably be overcome by pursuing such variations on the same theme.

Electronic Music Distribution (EMD) is in demand of robust, automatically extracted music descriptors. We introduce a timbral similarity measures for comparing music titles. This measure is based on a Gaussian model of cepstrum coefficients. We describe the timbre extractor and the corresponding timbral similarity relation. We describe experiments in assessing the quality of the similarity relation, and show that the measure is able to yield interesting similarity relations, in particular when used in conjunction with other similarity relations. We illustrate the use of the descriptor in several EMD applications developed in the context of the Cuidado European project.

Building highly concurrent systems, such as large-scale Internet services, requires managing many information flows at once and maintaining peak throughput when demand exceeds resource availability. In addition, any platform supporting Internet services must provide high availability and be able to cope with burstiness of load. Many approaches to building concurrent systems have been proposed, which generally fall into the two categories of threaded and event-driven programming. We propose that threads and events are actually on the ends of a design spectrum, and that the best implementation strategy for these applications is somewhere in between. We present a general-purpose design framework for building highly concurrent systems, based on three design components --- tasks, queues, and thread pools --- which encapsulate the concurrency, performance, fault isolation, and software engineering benefits of both threads and events. We present a set of design patterns that can be applied to...

We advocate the need for integrating semantic information into a file system. We demonstrate the benefits of this in Sedar, a deep archival file system. Sedar is the first archival file system that integrates semantic storage and retrieval capabilities. In addition, Sedar introduces several novel features: the notion of semantic-hashing to reduce the storage consumption that is robust against misalignment of documents; virtual snapshots of the namespace, and conceptual deletions of files and directories. Sedar exposes a semantic catalogue that allows other semanticbased tools (e.g., visualization and statistical analysis) to be built. It uses a decentralized P2P storage utility enabling horizontal scalability.

We present a system for indexing of asd index-based seas-ch in PCM-based audio material. Given a short excerpt of a waveform signal as a query, the index returns all pieces in a data base containing that waveform. Additionally, the precise position of the waveform within those pieces is returned. The indexing method is robust against several signal processing operations such as lossy compression or asalog trassmissions. Indexing of a test data base consisting of approximately 50 GB of audio data results in ash index of size 58 MB. Response times to queries of lengths of about one or a half of a second as-e only fractions of a second.

We propose a new design for highly-concurrent server applications such as Internet services. This design, the staged event-driven architecture (SEDA), is intended to support massive concurrency demands for a wide range of applications. In SEDA, applications are constructed as a set of event-driven stages separated by queues. This design allows services to be well-conditioned to load, preventing resources from being overcommitted when demand exceeds service capacity. Decomposing services into a set of stages enables modularity and code reuse, as well as the development of debugging tools for complex event-driven applications. We present the SEDA design, as well as Sandstorm, an Internet services platform based on this architecture. We evaluate the use of Sandstorm through two applications: A simple HTTP server benchmark and a packet router for the Gnutella peer-to-peer file sharing network. 1 Introduction The Internet presents a systems problem of unprecedented scale: that of supporti...

Because of the increasing number of music distributors offering an ever growing number of albums and tracks on the Internet, access methods such as retrieval, interactive exploration or similaritybased search demand more sophisticated technologies than metadatabased approaches currently o#er including queries for artists, albums, titles or manually assigned genre information. This requirement also holds for private collections with the tracks being more and more often stored uniformly in a library of, e.g., MP3 or OGG files as opposed to just a few tracks stored on separate media such as vinyl records, tapes or compact discs that have to be changed when playing more than one album.