GoodSounds dataset contains around 28 hours of recordings of single notes and scales played by 15 different professional musicians, all of them holding a music degree and having some expertise in teaching. 12 different instruments (flute, cello, clarinet, trumpet, violin, alto sax alto, tenor sax, baritone sax, soprano sax, oboe, piccolo and bass) were recorded using one or up to 4 different microphones. For all the instruments the whole set of playable semitones in the instrument is recorded several times with different tonal characteristics. Each note is recorded into a separate monophonic audio file of 48kHz and 32 bits. Rich annotations of the recordings are available, including details on recording environment and rating on tonal qualities of the sound (“good-sound”, “bad”, “scale-good”, “scale-bad”).

General description:

This dataset was created in the context of the Pablo project, partially funded by KORG Inc. It contains monophonic recordings of two kind of exercises: single notes and scales. The dataset was reported in the following article:

Romaní Picas O., Parra Rodriguez H., Dabiri D., Tokuda H., Hariya W., Oishi K., & Serra X."A real-time system for measuring sound goodness in instrumental sounds", 138th Audio Engineering Society Convention (2015).

The recordings were made in the Universitat Pompeu Fabra / Phonos recording studio by 15 different professional musicians, all of them holding a music degree and having some expertise in teaching. 12 different instruments were recorded using one or up to 4 different microphones (depending on the recording session). For all the instruments the whole set of playable semitones in the instrument is recorded several times with different tonal characteristics. Each note is recorded into a separate mono .flac audio file of 48kHz and 32 bits. The tonal characteristics are explained both in the the following section and the related publication.

The audio files are organised in one directory for each recording session. In addition to the files, one SQLite database file is included. The structure of the database is related in the following section.

Database description:

The database is meant for organizing the sounds in a handy way. It is organised in four different tables: sounds, takes, packs and ratings.

Sounds

The table containing the sounds annotations.

id

instrument : flute, cello, clarinet, trumpet, violin, sax_alto, sax_tenor, sax_baritone, sax_soprano, oboe, piccolo, bass

note

octave

dynamics : for some sounds, the musical notation of the loudness level (p, mf, f..)

recorded_at : recording date and time

location : recording place

player : the musician who recorded it. For detailed information about the musicians please contact us.

bow_velocity : for some string instruments the velocity of the bow (slow, medieum, fast)

bridge_position : for some string instruments the position of the bow (tasto, middle, ponticello)

string : for some string instruments the number of the string in which the sound it's played (1: lowest in pitch)

csv_file : used for creation of the DB

csv_id : used for creation of the DB

pack_filename : used for creation of the DB

pack_id : used for creation of the DB

attack : for single notes, manual annotation of the onset in samples.

decay : for single notes, manual annotation of the decay in samples.

sustain : for single notes, manual annotation of the beginnig of the sustained part in samples.

release : for single notes, manual annotation of the beginnig of the release part in samples.

offset : for single notes, manual annotation of the offset in samples

reference : 1 if sound was used to create the models in the good-sounds project, 0 if not.

klass : user generated tags of the tonal qualities of the sound. They also contain information about the exercise, that could be single note or scale.

"good-sound": good examples of single note

"bad": bad example of one of the sound attributes defined in the project (please read the papers for a detailed explanation)

"scale-good": good example of a one octave scale going up and down (15 notes). If the scale is minor a tagged "minor" is also available.

"scale-bad": bad example scale of one of the sounds defined in the project. (15 notes up and down).

comments : if any

semitone : midi note

pitch_reference : the reference pitch

Takes

A sound can have several takes as some of them were recorded using different microphones at the same time. Each take has an associated audio file.

id

microphone

filename : the name of the associated audio file

original_filename :

freesound_id : for some sounds uploaded to freesound.org

sound_id : the id of the sound in the DB

goodsound_id : for some of the sounds available in good-sounds.org

Packs

A pack is a group of sounds from the same recording session. The audio files are organised in the sound_files directory in subfolders with the pack name to which they belong.

id

name

description

Ratings

Some musicians rated some sounds in a 0-10 goodness scale for the user evaluatio of the first project prototype. Please read the paper for more detailed information.

id

mark: the rate or score.

type: the klass of the sound. Related to the tags of the sound.

created_at

comments

sound_id

rater: the musician who rated the sound.

License:

This work is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc/4.0/ or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA.

The Good-sounds dataset includes monophonic audio recordings of two types of musical exercises: single notes and scales. These recordings, performed by 15 professional musicians on flute, clarinet, trumpet, violin, and cello, were made in studio conditions at the Universitat Pompeu Fabra. The dataset supports research in sound quality, music education, and audio classification. It includes audio files in .wav format and metadata in .csv format detailing the exercise type, note, pitch, instrument, and performer, among others.

This dataset contains the ground truth data used to evaluate the musical pitch, tempo and key estimation algorithms developed during the AudioCommons H2020 EU project and which are part of the Audio Commons Audio Extractor tool. It also includes ground truth information for the single-eventness audio descriptor also developed for the same tool. This ground truth data has been used to generate the following documents: Deliverable D4.4: Evaluation report on the first prototype tool for the automatic semantic description of music samples Deliverable D4.10: Evaluation report on the second prototype tool for the automatic semantic description of music samples Deliverable D4.12: Release of tool for the automatic semantic description of music samples All these documents are available in the materials section of the AudioCommons website. All ground truth data in this repository is provided in the form of CSV files. Each CSV file corresponds to one of the individual datasets used in one or more evaluation tasks of the aforementioned deliverables. This repository does not include the audio files of each individual dataset, but includes references to the audio files. The following paragraphs describe the structure of the CSV files and give some notes about how to obtain the audio files in case these would be needed. Structure of the CSV files All CSV files in this repository (with the sole exception of SINGLE EVENT - Ground Truth.csv) feature the following 5 columns: Audio reference: reference to the corresponding audio file. This will either be a string withe the filename, or the Freesound ID (for one dataset based on Freesound content). See below for details about how to obtain those files. Audio reference type: will be one of Filename or Freesound ID, and specifies how the previous column should be interpreted. Key annotation: tonality information as a string with the form "RootNote minor/major". Audio files with no ground truth annotation for tonality are left blank. Ground truth annotations are parsed from the original data source as described in the text of deliverables D4.4 and D4.10. Tempo annotation: tempo information as an integer representing beats per minute. Audio files with no ground truth annotation for tempo are left blank. Ground truth annotations are parsed from the original data source as described in the text of deliverables D4.4 and D4.10. Note that integer values are used here because we only have tempo annotations for music loops which typically only feature integer tempo values. Pitch annotation: pitch information as an integer representing the MIDI note number corresponding to annotated pitch's frequency. Audio files with no ground truth pitch for tempo are left blank. Ground truth annotations are parsed from the original data source as described in the text of deliverables D4.4 and D4.10. The remaining CSV file, SINGLE EVENT - Ground Truth.csv, has only the following 2 columns: Freesound ID: sound ID used in Freesound to identify the audio clip. Single Event: boolean indicating whether the corresponding sound is considered to be a single event or not. Single event annotations were collected by the authors of the deliverables as described in deliverable D4.10. How to get the audio data In this section we provide some notes about how to obtain the audio files corresponding to the ground truth annotations provided here. Note that due to licensing restrictions we are not allowed to re-distribute the audio data corresponding to most of these ground truth annotations. Apple Loops (APPL): This dataset includes some of the music loops included in Apple's music software such as Logic or GarageBand. Access to these loops requires owning a license for the software. Detailed instructions about how to set up this dataset are provided here. Carlos Vaquero Instruments Dataset (CVAQ): This dataset includes single instrument recordings carried out by Carlos Vaquero as part of this master thesis. Sounds are available as Freesound packs and can be downloaded at this page: https://freesound.org/people/Carlos_Vaquero/packs Freesound Loops 4k (FSL4): This dataset set includes a selection of music loops taken from Freesound. Detailed instructions about how to set up this dataset are provided here. Giant Steps Key Dataset (GSKY): This dataset includes a selection of previews from Beatport annotated by key. Audio and original annotations available here. Good-sounds Dataset (GSND): This dataset contains monophonic recordings of instrument samples. Full description, original annotations and audio are available here. University of IOWA Musical Instrument Samples (IOWA): This dataset was created by the Electronic Music Studios of the University of IOWA and contains recordings of instrument samples. The dataset is available upon request by visiting this website. Mixcraft Loops (MIXL): This dataset includes some of the music loops included in Acoustica's Mixcraft music software. Access to thes...

vibrato_analysis

Vibrato Analysis Dataset,Detection and Parameterization

Introduction

This dataset was created as an outcome of a summer internship with Good-sounds project. Sound recordings were used for vibrato analysis. The dataset contains monophonic recordings of single notes and melodies for four instruments with annotations of four different recording sets regarding some vibrato parameters.

It is organized according to the owner of the recordings. Sounds with vibrato and no vibrato are presented within the folders with their annotations in cvs format. All annotations except than the alto saxophone recordings include derivative analysis of pitch parameters.

One of the recordings for violin was made in the Universitat Pompeu Fabra recording studio by a violist from MTG. Single notes of two octaves and four different melodies starting from different pitch are recorded in wav format sampled at 44.1 kHz. Each sound separated by semitones is recorded four times for no vibrato,slow, standard and fast rates of vibrato. Melodies were played for no vibrato and vibrato at a standard rate to have the attenuation at the end of the note.

One other recording set was taken from Good-sounds project. The only brass instrument within the whole dataset is this one for alto saxophone. It was again recorded in the Universitat Pompeu Fabra / Phonos recording studio. This dataset for alto saxophone not only contains vibrato and non vibrato sounds but also some different tonal characteristics.

Three other recordings are downloaded from a user named Carlos_Vaquero in Freesound. Violin, violoncello and transverse flute recordings were downloaded and used for non commercial purposes under creative commons license. Within the Carlos_Vaquero folder, audio files are separated according to the type of instruments.

Dataset Description

The database is meant for organizing the sounds in a handy way. It is organized according to the creator. In each three datasets, annotations and analysis parameters are available within the csv files and each has 11 field descriptor.

• Carlos_Vaquero
  • Transverse_flute
  • Violin
  • Violoncello
• Good-sounds-Alto sax
  • Alto-sax
• MTG - Violin
  • Violin

Except than alto-sax recordings, each contains following parameters:

• Peak_Percentage: Percent wise proportion of peaks in first derivation of the pitch trajectory.
• Mean_Difference: Mean value of the differences of two consecutive peaks.
• Max_Difference: Maximum separation of side by side peaks.
• Index_Max (first one): Index value of the maximum peak in the derivative array.
• Location_Max (%): Location of the maximum peak in the array, percent wise.
• Start_End_Time (seconds): Starting and ending time instants of vibrato in the recording.
• Duration: Duration of the vibrato part of the recording.

Documentation

Related report can be found in the GitHub repository as "Vibrato Analysis Internship Report".

License

All the software is distributed with the Affero GPL v3 license.

Vibrato Dataset is the extended version of the previous dataset. It consists of vibrato sounds for five different instruments providing different pitch ranges of notes, melodies, articulations, scales and musical techniques as staccato and legato. It includes 295 note tracks with vibratos and corresponding non vibrato tones and 49 melody and articulation tracks.

Data Collection

It is organized according to the source of the recordings. Sounds with vibrato and no vibrato are presented within the folders with their annotations in cvs format. All annotations except than the alto saxophone recordings include derivative analysis of pitch parameters.

The reason of choosing following subsets is that all subsets are created in MTG and they are open source.

The MTG violin recordings includes single notes of two octaves and four different melodies starting from different pitch are recorded in wav format sampled at 44.1 kHz. Each sound separated by semitones is recorded four times for no vibrato,slow, standard and fast rates of vibrato. Melodies were played for no vibrato and vibrato at a standard rate to have the attenuation at the end of the note. Good-sounds alto-sax recordings main feature is that it offers the quality of the vibrato sound and provides vibrato tracks of articulation, scales and staccato, legato techniques. The remaining subset offers a good variety of instruments and mostly two octave range of vibrato-non vibrato pairs.

Dataset Description

The database is meant for organizing the sounds in a handy way. It is organized according to the creator. In each three datasets, annotations and analysis parameters are available within the csv files and each has 11 field descriptor.

Carlos_Vaquero

Transverse_flute

Violin

Violoncello

Bassoon

Good-sounds-Alto sax

Alto-sax

MTG - Violin

Violin

Except than alto-sax recordings, each annotation files contain following parameters:

Peak_Percentage: Percent wise proportion of peaks in first derivation of the pitch trajectory.

Mean_Difference: Mean value of the differences of two consecutive peaks.

Max_Difference: Maximum separation of side by side peaks.

Index_Max (first one): Index value of the maximum peak in the derivative array.

Location_Max (%): Location of the maximum peak in the array, percent wise.

Start_End_Time (seconds): Starting and ending time instants of vibrato in the recording.

Duration: Duration of the vibrato part of the recording.

License

All the software is distributed with the Affero GPL v3 license.