This dataset contains several hours of recordings of Hardanger fiddle music, from Norway, with note annotations of onsets, offsets and pitches, provided by the performers themselves. A subset has also been annotated with beat onset positions by the performer as well as three expert musicians. Beat annotation is often recorded as positions in seconds, without explicit connection with actual musical events. Here, instead, beat onsets are associated with the onsets of the notes that represent the start of each beat. The motivation underlying the chosen approach is discussed in [our TISMIR paper][1]. This work has been carried out in the context of the [MIRAGE project][2]. The recording and subsequent note annotations were made by three musicians: two music students, skilled fiddlers, from the Norwegian Academy of Music – Henrik Nordtun Gjertsen (S1) and Astrid Garmo (S2) – and one renowned fiddler: [Olav Luksengård Mjelva][3] (P). Update March 2025: 12 new tunes (Volume II) played by Olav Luksengård Mjelva. The careful correction of the note annotations by the musician was supported by an additional funding provided by the [RITMO Centre][4]. ## Audio recordings ## The audio recordings are in stereo, in WAV format, with a sampling rate of 44100 Hz and a bit depth of 16 bits. P’s recordings were carried out in a studio in an old log building with natural “wooden house acoustics”, using WA84 stereo mics, U67 replica room mics, an Audient asp880 preamp and the Presonus Firepod interface. S1’s recordings were carried out in a room with wooden walls, with relatively normal acoustics, using a Zoom H6 recorder. S2’s recordings were carried out in a small and relatively dry room using a Zoom recorder. The musicians were not specifically asked to tune their fiddle to a specific diapason, leaving them to proceed as more natural for them. For each tune, five different versions were recorded: first playing in a normal way, and then following four distinct expressions: sad, angry, happy, and tender. The normal version of each song is stored in a WAV file with the tune title (e.g., Haslebuskane.wav). For the four variants, the WAV file name adds also the expression type (e.g., Haslebuskane_happy.wav). S1 and S2's recordings are also available in a version remastered by Diana Gorovaya at Saturday Mastering. They can be found in the "mastered" subfolder in each tune folder. ## Note annotation ## We ask three musicians to record tunes they are familiar with, and to annotate notes using computer assistance tools as aid. In this way, they could base the transcription not only on what they heard from the recordings, but also on a memory of what they actually played. In order to avoid any bias, instead of offering the annotators the possibility to correct computer-generated annotations, the annotation (of the normal version, as explained below) is carried out entirely from scratch. For each tune, the normal version is first annotated by the musician; then the note annotations are automatically transferred to the other versions, and further checked and corrected by the musician (Elowsson and Lartillot, 2021). We encouraged performers to be very careful regarding onsets, and try to keep errors within 20 ms. This means that we can only allow a very narrow margin of error for the annotations to ensure that they can be reliably used for evaluation. For each tune, the audio recording of the normal version has a correspond note annotation as a CSV file, with the same file name (e.g., Haslebuskane.csv). In the CSV files, each line corresponds to a note, and the columns "onset" and "offset" indicate the start and end time, in seconds, of the notes, while "onpitch" indicates the pitch (using the MIDI convention, with 60 = middle C, and A4 arbitrarily tuned at 440 Hz, but using real values to indicate intonations in details). In the column "essentials", most notes have value 1, but those with value 0 correspond to notes flagged by the annotator as ornamental. For each normal version, an additional MAT file contains the Pitchogram stored in Matlab format, as explained in the TISMIR paper. The file can be used with the Annotemus software. ## Beat annotation ## The dataset contains the beat annotations for the 12 tunes (totalling 18 minutes) played by musician P. To study the degree of agreement among music experts concerning beat annotations, each tune has been annotated by three experts in addition to the musician (P): - a Scandinavian folk music scholar and fiddle music expert, Mats Sigvard Johansson (M), - two music students from the University of South- Eastern Norway – Guro Kvifte Nesheim (S3) and Mattias Cyvin (S4), experts in Hardanger fiddle music. We ask the annotators to annotate beat onsets by selecting particular notes in the annotations. In other words, the onset of each chosen note would correspond to the onset of each successive (non-silent) beat. In the case several notes are played synchronously at a given beat onset, it does not matter which note is selected, as we do not find it necessary, at least for the current study, to address such level of temporal precision. Annotators have been asked to skip silent beats, but with the possibility to annotate beat subdivisions at the vicinity of the silent beats. Cf. example in Figure 3 of the TISMIR paper. For each normal version of each tune played by the musician P, are associated 5 CSV files, corresponding to the annotations by the 5 annotators (e.g., Fuglesangen_P.csv, Fuglesangen_M.csv, etc.). In each CSV file, the beats are indicated by a non-zero value in the "bar" column (indicating the bar number related to that beat), as well as a value in the "upmeter" column (indicating the beat positions in that bar, either 1, 2 or 3). These experts were subsequently asked to compare their annotation with the version by the musician. This enables to obtain beat annotations for which the variability of the expert (with respect to the musician) has been fully reflected by the expert themselves, so that divergence can be considered as valid alternative beat grids. One of the annotators, S4, compared all the beat annotations of each tune, combining the multiple versions of beat annotations of each tune into one single authoritative version, as explained in the TISMIR paper. This corresponds to the CSV file without suffix (e.g., Fuglesangen.csv). ## Acknowledgements ## We warmly thank the musicians who recorded the dataset: Olav Luksengård Mjelva (P), Henrik Nordtun Gjertsen (S1) and Astrid Garmo (S2). The beat onsets were annotated by Mats Sigvard Johansson (M), Guro Kvifte Nesheim (S3) and Mattias Cyvin (S4). We thank Hans-Hinrich Thedens from the National Library of Norway, and Unni Løvlid from the Norwegian Academy of Music for advising for this project. This work is supported by the Research Council of Norway through its Centers of Excellence scheme, project number 262762 and the [MIRAGE project][1], grant number 287152. [1]: https://transactions.ismir.net/articles/10.5334/tismir.139 [2]: https://www.uio.no/ritmo/english/projects/mirage/.no/ritmo/english/projects/mirage/. [3]: https://www.olavmjelva.no [4]: https://www.uio.no/ritmo/english/