Combining melody processing and information retrieval techniques: Methodology, evaluation, and system implementation

Choral singing is a central part of musical cultures across the world, yet many facets of this widespread form of polyphonic singing are still to be explored. Music information retrieval (MIR) research on choral singing benefits from multitrack recordings of the individual singing voices. However, there exist only few publicly available multitrack datasets on polyphonic singing. In this paper, we present Dagstuhl ChoirSet (DCS), a multitrack dataset of a cappella choral music designed to support MIR research on choral singing. The dataset includes recordings of an amateur vocal ensemble performing two choir pieces in full choir and quartet settings. The audio data was recorded during an MIR seminar at Schloss Dagstuhl using different close-up microphones to capture the individual singers’ voices. In this article, we give detailed insights into all stages of creating DCS: recording process, data preparation, generation of annotations as well as development of suitable interfaces for publicly accessing and reusing the data. Furthermore, we demonstrate the potential of the dataset for MIR research by discussing case studies on choral intonation assessment and multiple fundamental frequency (F0) estimation.

Musical themes are essential elements in Western classical music. In this paper, we present the Musical Theme Dataset (MTD), a multimodal dataset inspired by “A Dictionary of Musical Themes” by Barlow and Morgenstern from 1948. For a subset of 2067 themes of the printed book, we created several digital representations of the musical themes. Beyond graphical sheet music, we provide symbolic music encodings, audio snippets of music recordings, alignments between the symbolic and audio representations, as well as detailed metadata on the composer, work, recording, and musical characteristics of the themes. In addition to the data, we also make several parsers and web-based interfaces available to access and explore the different modalities and their relations through visualizations and sonifications. These interfaces also include computational tools, bridging the gap between the original dictionary and music information retrieval (MIR) research. The dataset is of relevance for various subfields and tasks in MIR, such as cross-modal music retrieval, music alignment, optical music recognition, music transcription, and computational musicology.

Representing a significant shift from methods that employ bibliographic metadata for music retrieval, modern Music Information Retrieval (MIR) research seeks to utilize characteristics of music to search databases of musical content via musically expressed queries. These investigations hold significant promise for the future development of commercial and research applications such as searchable databases of popular tunes on the web and databases of historical music. However the challenges are diverse, encompassing music cognition, storage and processing, and traditional information retrieval (IR) issues. This paper focuses on IR issues in MIR, discussing various approaches to music representation and matching techniques. Prototype MIR systems Meldex, Semex, Themefinder and Arthur are briefly described.

Most Music Information Retrieval (MIR) researchers will agree that understanding users' needs and behaviors is critical for developing a good MIR system. The number of user studies in the MIR domain has been gradually increasing since the early 2000s, reflecting this growing appreciation of the need for empirical studies of users. However, despite the growing number of user studies and the wide recognition of their importance, it is unclear how great their impact has been in the field: on how systems are developed, how evaluation tasks are created, and how MIR system developers in particular understand critical concepts such as music similarity or music mood. In this paper, we present our analysis on the growth, publication and citation patterns, topics, and design of 198 user studies. This is followed by a discussion of a number of issues/challenges in conducting MIR user studies and distributing the research results. We conclude by making recommendations to increase the visibility and impact of user studies in the field.

Audio and music generation systems have been remarkably developed in the music information retrieval (MIR) research field. The advancement of these technologies raises copyright concerns, as ownership and authorship of AI-generated music (AIGM) remain unclear. Also, it can be difficult to determine whether a piece was generated by AI or composed by humans clearly. To address these challenges, we aim to improve the accuracy of AIGM detection by analyzing the structural patterns of music segments. Specifically, to extract musical features from short audio clips, we integrated various pre-trained models, including self-supervised learning (SSL) models or an audio effect encoder, each within our suggested transformerbased framework. Furthermore, for long audio, we developed a segment transformer that divides music into segments and learns inter-segment relationships. We used the FakeMusicCaps and SONICS datasets, achieving high accuracy in both the short-audio and full-audio detection experiments. These findings suggest that integrating segment-level musical features into long-range temporal analysis can effectively enhance both the performance and robustness of AIGM detection systems.

This paper investigates the generic problem of model selection in the specific context of Music Information Retrieval (MIR). In MIR research, similarity measures are developed for ranking musical items with respect to their relevance to a user's musical query. The application of such similarity measures in MIR systems typically requires musical works to be divided into more manageable units. This involves two tasks: melody segmentation and voice separation. For both of these tasks, several computational models have been proposed in the symbolic domain. It seems reasonable to assume that those solutions that are most in accordance with human performance will result in the best ranking of retrieval output. We conducted two experiments, each with twenty experts and twenty novices. In the melody segmentation experiment, we found a high agreement between the participants. Evaluating algorithm output against participant data, we conclude that human output cannot be distinguished from three of the segmentation algorithms (Grouper, IDyOM and LBDM). For voice separation—which we evaluated by means of a melody identification task—the situation is different, as the combined results of two algorithms (Skyline and SSA) were shown to agree best with experimental results, and differences were found between novice and expert performance. Several other model selection criteria besides performance are discussed in conclusion.

Humans have used electronic information retrieval (IR) systems for more than 50 years as they evolved from experimental systems to full-scale Web search engines and digital libraries. The fields of library and information science (LIS), cognitive science, human factors and computer science have historically been the leading disciplines in conducting research that seeks to model human interaction with IR systems for all kinds of information related behaviors. As technology problems have been mastered, the theoretical and applied framework for studying human interaction with IR systems has evolved from systems-centered to more user-centered, or cognitive-centered approaches. However, cognitive information retrieval (CIR) research that focuses on user interaction with IR systems is still largely under-funded and is often not included at computing and systems design oriented conferences. But CIR-focused research continues, and there are signs that some IR systems designers in academia and the Web search business are realizing that user behavior research can provide valuable insights into systems design and evaluation. The goal of our book is to provide an overview of new CIR research directions. This book does not provide a history of the research field of CIR. Instead, the book confronts new ways of looking at the human information condition with regard to our increasing need to interact with IR systems. The need has grown due to a number of factors, including the increased importance of information to more people in this information age. Also, IR was once considered document-oriented, but has now evolved to include multimedia, text, and other information objects. As a result, IR systems and their complexity have proliferated as users and user purposes for using them have also proliferated. Human interaction with IR systems can often be frustrating as people often lack an understanding of IR system functionality. New more holistic directions in CIR are emerging that conceptualize human-IR system interaction at the human-computer interaction level (taking into account interface issues), the information behavior level (taking into account the role of IR system interaction in the total range of people’s information behaviors), and the

Purpose Prior Information Retrieval (IR) research synthesizes progress from individual studies, yet academia-industry collaboration dynamics remain unexplored. This study investigates: (1) productivity patterns and venues, (2) citations-downloads relationships, (3) topic evolution, and (4) collaboration trends. Design/methodology/approach We perform an analysis of 53,471 ACM IR papers (2000–2018) using bibliometrics and DistilBERT topic modeling. Findings We find that industry-involved papers preferred WWW/CIKM venues; collaborations dominated RecSys/CSCW. We see that academia-industry collaborations achieved the highest download-to-citation conversion rates. Academia focused on algorithms; industry on applications; collaborations bridged both with rising human-centered themes. Research implications This is a pioneering large-scale bibliometrics revealing collaboration’s impact on IR knowledge evolution and provides a methodological framework for cross-sector analysis. Practical implications The paper identifies optimal venues (RecSys/CSCW) for partnerships and guides joint initiatives (shared datasets, grants) to bridge academia-industry divides and enhance research translation. Originality/value This is the first large-scale bibliometric analysis of IR academia-industry collaboration. The paper finds many novel insights, including the fact that collaboration boosts citation efficiency, enables complementary specialization, and drives topic convergence.

Towards improving the performance in various music information processing tasks, recent studies exploit different modalities able to capture diverse aspects of music. Such modalities include audio recordings, symbolic music scores, mid-level representations, motion, and gestural data, video recordings, editorial or cultural tags, lyrics and album cover arts. This paper critically reviews the various approaches adopted in Music Information Processing and Retrieval and highlights how multimodal algorithms can help Music Computing applications. First, we categorize the related literature based on the application they address. Subsequently, we analyze existing information fusion approaches, and we conclude with the set of challenges that Music Information Retrieval and Sound and Music Computing research communities should focus in the next years.

We present the open-source implementation of the first fully automatic and comprehensive DJ system, able to generate seamless music mixes using songs from a given library much like a human DJ does.The proposed system is built on top of several enhanced music information retrieval (MIR) techniques, such as for beat tracking, downbeat tracking, and structural segmentation, to obtain an understanding of the musical structure. Leveraging the understanding of the music tracks offered by these state-of-the-art MIR techniques, the proposed system surpasses existing automatic DJ systems both in accuracy and completeness. To the best of our knowledge, it is the first fully integrated solution that takes all basic DJing best practices into account, from beat and downbeat matching to identification of suitable cue points, determining a suitable cross-fade profile and compiling an interesting playlist that trades off innovation with continuity.To make this possible, we focused on one specific sub-genre of electronic dance music, namely Drum and Bass. This allowed us to exploit genre-specific properties, resulting in a more robust performance and tailored mixing behavior.Evaluation on a corpus of 160 Drum and Bass songs and an additional hold-out set of 220 songs shows that the used MIR algorithms can annotate 91% of the songs with fully correct annotations (tempo, beats, downbeats, and structure for cue points). On these songs, the proposed song selection process and the implemented DJing techniques enable the system to generate mixes of high quality, as confirmed by a subjective user test in which 18 Drum and Bass fans participated.

Within the broad problem known as automatic music transcription, we considered the specific task of automatic drum transcription (ADT). This is a complex task that has recently shown significant advances thanks to deep learning (DL) techniques. Most notably, massive amounts of labeled data obtained from crowds of annotators have made it possible to implement large-scale supervised learning architectures for ADT. In this study, we explored the untapped potential of these new datasets by addressing three key points: First, we reviewed recent trends in DL architectures and focused on two techniques, self-attention mechanisms and tatum-synchronous convolutions. Then, to mitigate the noise and bias that are inherent in crowdsourced data, we extended the training data with additional annotations. Finally, to quantify the potential of the data, we compared many training scenarios by combining up to six different datasets, including zero-shot evaluations. Our findings revealed that crowdsourced datasets outperform previously utilized datasets, and regardless of the DL architecture employed, they are sufficient in size and quality to train accurate models. By fully exploiting this data source, our models produced high-quality drum transcriptions, achieving state-of-the-art results. Thanks to this accuracy, our work can be more successfully used by musicians (e.g., to learn new musical pieces by reading, or to convert their performances to MIDI) and researchers in music information retrieval (e.g., to retrieve information from the notes instead of audio, such as the rhythm or structure of a piece).

Due to lack of mature techniques in privacy-preserving information retrieval (IR), concerns about information privacy and security have become serious obstacles that prevent valuable user data to be used in IR research such as studies on query logs, social media, and medical record retrieval. In SIGIR 2014 and SIGIR 2015, we have run the privacy-preserving IR workshops exploring and understanding the privacy and security risks in information retrieval. This year, we continue the efforts of connecting the two disciplines of IR and privacy/security by organizing this workshop. We target on three themes, differential privacy and IR dataset release, privacy in search and browsing, and privacy in social media. The workshop includes panels with researchers from both fields on these three themes, as well as invite industry speakers for real-world challenges. The goals of this workshop include (1) bringing together the two research fields, and (2) yielding fruitful collaborations.

Information retrieval (IR) and information privacy/security are two fast-growing computer science disciplines. There are many synergies and connections between these two disciplines. However, there have been very limited efforts to connect the two important disciplines. On the other hand, due to lack of mature techniques in privacy-preserving IR, concerns about information privacy and security have become serious obstacles that prevent valuable user data to be used in IR research such as studies on query logs, social media, tweets, sessions, and medical record retrieval. This privacy-preserving IR workshop aims to spur research that brings together the research fields of IR and privacy/security, and research that mitigates privacy threats in information retrieval by constructing novel algorithms and tools that enable web users to better understand associated privacy risks.

The Web has become a large repository of documents (or pages) written in many different languages. In this context, traditional information retrieval (IR) techniques cannot be used whenever the user query and the documents being retrieved are in different languages. To address this problem, new cross‐language information retrieval (CLIR) techniques have been proposed. In this work, we describe a method for cross‐language retrieval of medical information. This method combines query terms and related medical concepts obtained automatically through a categorization procedure. The medical concepts are used to create a linguistic abstraction that allows retrieval of information in a language‐independent way, minimizing linguistic problems such as polysemy. To evaluate our method, we carried out experiments using the OHSUMED test collection, whose documents are written in English, with queries expressed in Portuguese, Spanish, and French. The results indicate that our cross‐language retrieval method is as effective as a standard vector space model algorithm operating on queries and documents in the same language. Further, our results are better than previous results in the literature.

Welcome to the 40th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2017). SIGIR is the premier scientific conference in the broad area of information retrieval. We received a high-quality set of full-paper submissions to consider for inclusion in the conference program. We thank the 66 Senior Program Committee (SPC) members, 216 Program Committee (PC) members, and at least 70 additional reviewers for their contributions to paper selection. This pool of committed SIGIR volunteers was based in 33 countries and over 170 institutions, spanning academia, industry, and beyond. We recognized 8 SPC members and 29 PC members / additional reviewers whose contributions to the review process were outstanding. We reviewed 362 papers for the full paper track, and accepted 78, for an acceptance rate of 22%. In total, there were 1056 authors from 42 countries: 299 from China, 232 from the United States, and 525 from the rest of the world. The countries (with at least 10 submissions) with the highest acceptance rates were the United Kingdom (42%), Italy (42%), Singapore (38%), Australia (27%), and the United States (24%). The most popular call-for-papers topics for accepted papers were ranking algorithms (19%), text representation (17%), behavioral analysis (15%), recommender systems (14%), and user studies (14%). We made several modifications to the paper submission and review process for SIGIR 2017. We transitioned to EasyChair, one of three conference submission applications supported by ACM. We significantly updated the call for papers to broaden the range of topics considered in scope for the conference, and more accurately reflect current research trends. To boost diversity, we increased the number of women SPC members (24% this year, up from 15% last year) and increased the number of SPC members based in Asia (23% this year, up from 8% last year). To help improve review quality, we also required that PC members have at least one full paper at SIGIR and be active at the SIGIR conference in the past five years (some limited exceptions were permitted). The largest change was a modification to the role of the secondary SPC, with the goal of better utilizing their expertise and making better decisions for borderline papers. Each paper had a primary SPC member, who was responsible for writing the metareview and making a recommendation regarding the paper's disposition. If they believed it was necessary, the primary SPC could request an additional review from a secondary SPC member, who did not see the other reviews or the discussion to date. Once they submitted the additional review, the secondary SPC member could see the other reviews and participate in the discussion. The additional SPC review informed discussion and helped the primary SPC make a recommendation. Overall, secondary SPC reviews were provided for 20% of full papers. We ran an in-person SPC meeting at Microsoft in Redmond, WA. To allow more SPC members to take part, we enabled remote participation via video conference. Approximately half of the SPC participated, either in person or remotely. SPC members were asked to submit their full-paper recommendations prior to the meeting. During the meeting, we discussed papers flagged by SPCs as requiring discussion, as well as papers with "accept" and "reject" recommendations. Some of the papers marked as "reject" by SPC members before the meeting had strong support from at least one PC member. A subset of those full papers ended up being included in the conference program after a detailed discussion. The short paper track received a record number of 398 submissions and accepted 121 (30%) of them. The track was chaired by Hui Fang, Mounia Lalmas and Yiqun Liu. The demo paper track received 36 submissions and accepted 17 (47%) of them. The track was chaired by Gareth Jones and Takehiro Yamamoto. The SIGIR Symposium on IR in Practice (SIRIP 2017) had seven invited talks and five accepted regular papers. The track was chaired by Sumio Fujita and Vanessa Murdock. The doctoral consortium track received 14 submissions of which 9 (64%) were accepted. The track was chaired by Makoto P. Kato and Kira Radinsky. The tutorial track received 14 proposals of which 8 (57%) were accepted. The track was chaired by Fernando Diaz and Adam Jatowt. The workshop track received 12 proposals of which 8 (67%) were accepted. The track was chaired by Jaap Kamps and Grace Hui Yang.