2021:Lyrics Transcription (former: Automatic Lyrics-to-Audio Alignment)

Description

This year we host the Automatic Lyrics Transcription challenge. You are free to participate in one of the tasks or both of them. The task of Lyrics Transcription aims to identify the words from sung music audio, in the same way as in automatic speech recognition.

The algorithm receives either monophonic singing performances or a polyphonic mix (singing voice + musical accompaniment). Ideally, the lyrics transcriber should output meaningful word sequences:

Datasets

Training Datasets

Datasets within automatic lyrics transcription research can be categorised under two domains: Monophonic and polyphonic recordings. The former is considered to have only one singer singing the lyrics, and the latter is when there is music accompaniment. In this challenge, we recommend using the open source datasets below:

DAMP dataset

The DAMP - Sing!300x30x2 dataset which consists of solo singing recordings (monophonic) performed by amateur singers, collected via a mobile Karaoke application. The data is curated to be gender-wise balanced and contains performers from 30 different countries, which introduces a good amount of variability in terms of accents and pronunciation. list of recordings. For more details see the paper.

• The audio can be downloaded from the Smule web site
• Lyrics boundary annotations can be generated from raw annotations using this repository.
• Or annotations can be directly retrieved in the Kaldi form here

DALI Dataset

DALI (a large Dataset of synchronised Audio, LyrIcs and notes) is the benchmark dataset for building an acoustic model on polyphonic recordings and it contains over 5000 songs with semi-automatically aligned lyrics annotations. The songs are commercial recordings in full-duration, whereas the lyrics are described according to different levels of granularity including words and notes (and syllables underlying a given note). For each song DALI provides a link to a matched youtube video, from which the audio could be retrieved.

• For more details how, see its full description here.

Evaluation Datasets

The following datasets are used for evaluation and so cannot be used by participants to train their models under any circumstance.

Hansen's Dataset

The dataset contains 9 pop music songs in English with annotations of both beginnings- and ending-timestamps of each word. The ending timestamps are for convenience (copies of next word's beginning timestamp) and are not used in the evaluation. Sentence-level annotations are also provided. The audio has two versions: the original mix with instrumental accompaniment and a cappella singing voice only one. An example song can be seen here

You can read in detail about how the dataset was made here: Recognition of Phonemes in A-cappella Recordings using Temporal Patterns and Mel Frequency Cepstral Coefficients. The dataset has been kindly provided by Jens Kofod Hansen.

• file duration up to 4:40 minutes (total time: 35:33 minutes)
• 3590 words annotated in total

Mauch's Dataset

The dataset contains 20 pop music songs in English with annotations of beginning-timestamps of each word. Non-vocal sections are not explicitly annotated (but remain included in the last preceding word). We prefer to leave it this way, to enable comparison to previous work, evaluated on this dataset. The audio has instrumental accompaniment. An example song can be seen here.

You can read in detail about how the dataset was used for the first time here: Integrating Additional Chord Information Into HMM-Based Lyrics-to-Audio Alignment. The dataset has been kindly provided by Sungkyun Chang.

• file duration up to 5:40 minutes (total time: 1h 19m)
• 5050 words annotated in total

Jamendo Dataset

This dataset contains 20 full-duration music pieces with 10 different Western music genres, annotated with start-of-word timestamps. All songs have instrumental accompaniment. It is available online on Github, although note that we do not allow tuning model parameters using this data, it can only be used to gain insight into the general structure of the test data. For more information also refer to this paper.

• file duration up to 4:43 (total time: 1h 12m)
• 5677 words annotated in total

Phonetization

A popular choice for phonetization of the words is the CMU pronunciation dictionary. One can phonetize them with the online tool. A list of all words of both datasets, which are outside of the list of CMU words is given here.

Audio Format

The data are sound wav/mp3 files, plus the associated word boundaries (in csv-like .txt/.tsv files)

• CD-quality (PCM, 16-bit, 44100 Hz)
• single channel (mono) for a cappella and two channels for original

Evaluation

Transcription

Word Error Rate (WER) - the standard metric use in Automatic Speech Recognition.

Alignment

The submitted algorithms will be evaluated at the word boundaries for the originally mixed songs (a cappella singing + instrumental accompaniment). Evaluation metrics on the a cappella singing can be reported as well on request, for the sake of getting insights on the impact of instrumental accompaniment on the algorithm, but will not be considered for the ranking.

Average absolute error/deviation Initially utilized in Mesaros and Virtanen (2008), the absolute error measures the time displacement between the actual timestamp and its estimate at the beginning and the end of each lyrical unit. The error is then averaged over all individual errors. An error in absolute terms has the drawback that the perception of an error with the same duration can be different depending on the tempo of the song. Here is a test of using this metric.

Percentage of correct segments The perceptual dependence on tempo is mitigated by measuring the percentage of the total length of the segments, labeled correctly to the total duration of the song. This metric is suggested by Fujihara et al. (2011), Figure 9. Here is a test of using this metric.

Percentage of correct estimates according to a tolerance window A metric that takes into consideration that the onset displacements from ground truth below a certain threshold could be tolerated by human listeners. We use 0.3 seconds as the tolerance window. This metric is suggested in Integrating Additional Chord Information Into HMM-Based Lyrics-to-Audio Alignment. Here is a test of using this metric.

For more detailed definition and formulas about the metrics, please check the section 2.2.1 of this thesis.

To obtain all three metrics for one detected output:

python eval.py <file path of the reference word boundaries> <file path of the detected word boundaries>

Note that evaluation scripts depend on mir_eval.

Submission Format

Submissions should be packaged and contain at least two files: The algorithm itself (as a binary or source code) and a README containing contact information and detailing, in full, the use of the algorithm.

Input Data

Participating algorithms will have to receive the following input format:

Transcription

• Audio in wav, 44.1kHz, stereo.

Alignment

• Audio in wav, 44.1kHz, stereo.
• Lyrics in .txt file where each word is separated by a space, each lyrics phrase is separated by a line break mark (\n).

Output File Format

Transcription

A list of words separated by white space <word1> <word2> ... Any non-word items (e.g. silence, end of the sentence) should be excluded.

Alignment

The alignment output file format is a tab-delimited ASCII text format.

Three column text file of the format

<onset_time(sec)>\t<offset_time(sec)>\t<label>\n
<onset_time(sec)>\t<offset_time(sec)>\t<label>\n
...

where \t denotes a tab, \n denotes the end of the line. The < and > characters are not included. An example output file would look something like:

0.000    5.223    word1
5.223    15.101   word2
15.101   20.334   word3

NOTE: the offset timestamps column is utilized only by the percentage of correct segments metric. Therefore skipping the second column is acceptable, and could result in degraded performance of this respective metric only.

Command line calling format

The submitted algorithm must take as arguments .wav file, .txt file as well as the full output path and filename of the output file. The ability to specify the output path and file name is essential. Denoting the input .wav file path and name as %input_audio; the lyrics .txt file as %input_txt and the output file path and name as %output, a program called foobar could be called from the command-line as follows:

foobar %input_audio (%input_txt) %output
foobar -i %input_audio (-it %input_txt)  -o %output

README File

A README file accompanying each submission should contain clear instructions on how to run the program (as well as contact information, etc.). In particular, each command line to run should be specified, using %input for the input sound file and %output for the resulting text file.

Submission closing dates

Closing date: TBD

Questions?

• send us an email - e.demirel@qmul.ac.uk (Emir Demirel) or info@voicemagix.com (Georgi Dzhambazov)

Potential Participants

Chitralekha Gupta

Emir Demirel

Gerardo Roa Dabike

Bibliography

Stoller, D. and Durand, S. and Ewert, S. (2019) End-to-end Lyrics Alignment for Polyphonic Music Using An Audio-to-Character Recognition Model. ICASSP 2019.

Sharma B, Gupta C. (2019) Automatic Lyrics-to-audio Alignment on Polyphonic Music Using Singing-adapted Acoustic Models. ICASSP 2019

Lee S. W., Scott, J. (2017) Word-level lyrics-audio synchronization using separated vocals", Acoustics Speech and Signal Processing, ICASSP IEEE International Conference on, pp. 646-650

Chang, S., & Lee, K. (2017). Lyrics-to-Audio Alignment by Unsupervised Discovery of Repetitive Patterns in Vowel Acoustics. arXiv preprint arXiv:1701.06078.

Pons, J. Gong, R. and Serra, X. (2017). Score-informed syllable segmentation for a cappella singing voice with convolutional neural networks. ISMIR 2017

Kruspe, A. (2016). Bootstrapping a System for Phoneme Recognition and Keyword Spotting in Unaccompanied Singing, ISMIR 2016

Dzhambazov, G. and Serra, X. (2015) Modeling of phoneme durations for alignment between polyphonic audio and lyrics, in 12th Sound and Music Computing Conference

Fujihara, H., & Goto, M. (2012). Lyrics-to-audio alignment and its application. In Dagstuhl Follow-Ups (Vol. 3). Schloss Dagstuhl-Leibniz-Zentrum fuer Informatik.

Mauch, M., Fujihara, H., & Goto, M. (2012). Integrating additional chord information into HMM-based lyrics-to-audio alignment. IEEE Transactions on Audio, Speech, and Language Processing, 20(1), 200-210.

Fujihara, H. Goto, M. Ogata, J. and Okuno, H. G. (2011) Lyricsynchronizer: Automatic synchronization system between musical audio signals and lyrics, IEEE Journal of Selected Topics in Signal Processing

Mesaros, A. and Virtanen, T. (2008), Automatic alignment of music audio and lyrics, in Proceedings of the 11th Int. Conference on Digital Audio Effects (DAFx-08), Espoo, Finland, 2008.