The LibriSpeech corpus is a collection of approximately 1,000 hours of audiobooks that are a part of the LibriVox project. Most of the audiobooks come from the Project Gutenberg. The training data is split into 3 partitions of 100hr, 360hr, and 500hr sets while the dev and test data are split into the ’clean’ and ’other’ categories, respectively, depending upon how well or challenging Automatic Speech Recognition systems would perform against. Each of the dev and test sets is around 5hr in audio length. This corpus also provides the n-gram language models and the corresponding texts excerpted from the Project Gutenberg books, which contain 803M tokens and 977K unique words.

1,732 PAPERS • 11 BENCHMARKS

VoxCeleb1 is an audio dataset containing over 100,000 utterances for 1,251 celebrities, extracted from videos uploaded to YouTube.

537 PAPERS • 9 BENCHMARKS

Multimodal Emotion Recognition IEMOCAP The IEMOCAP dataset consists of 151 videos of recorded dialogues, with 2 speakers per session for a total of 302 videos across the dataset. Each segment is annotated for the presence of 9 emotions (angry, excited, fear, sad, surprised, frustrated, happy, disappointed and neutral) as well as valence, arousal and dominance. The dataset is recorded across 5 sessions with 5 pairs of speakers.

534 PAPERS • 3 BENCHMARKS

Audioset is an audio event dataset, which consists of over 2M human-annotated 10-second video clips. These clips are collected from YouTube, therefore many of which are in poor-quality and contain multiple sound-sources. A hierarchical ontology of 632 event classes is employed to annotate these data, which means that the same sound could be annotated as different labels. For example, the sound of barking is annotated as Animal, Pets, and Dog. All the videos are split into Evaluation/Balanced-Train/Unbalanced-Train set.

523 PAPERS • 4 BENCHMARKS

The Universal Dependencies (UD) project seeks to develop cross-linguistically consistent treebank annotation of morphology and syntax for multiple languages. The first version of the dataset was released in 2015 and consisted of 10 treebanks over 10 languages. Version 2.7 released in 2020 consists of 183 treebanks over 104 languages. The annotation consists of UPOS (universal part-of-speech tags), XPOS (language-specific part-of-speech tags), Feats (universal morphological features), Lemmas, dependency heads and universal dependency labels.

491 PAPERS • 15 BENCHMARKS

VoxCeleb2 is a large scale speaker recognition dataset obtained automatically from open-source media. VoxCeleb2 consists of over a million utterances from over 6k speakers. Since the dataset is collected ‘in the wild’, the speech segments are corrupted with real world noise including laughter, cross-talk, channel effects, music and other sounds. The dataset is also multilingual, with speech from speakers of 145 different nationalities, covering a wide range of accents, ages, ethnicities and languages. The dataset is audio-visual, so is also useful for a number of other applications, for example – visual speech synthesis, speech separation, cross-modal transfer from face to voice or vice versa and training face recognition from video to complement existing face recognition datasets.

434 PAPERS • 5 BENCHMARKS

This CSTR VCTK Corpus includes speech data uttered by 110 English speakers with various accents. Each speaker reads out about 400 sentences, which were selected from a newspaper, the rainbow passage and an elicitation paragraph used for the speech accent archive. The newspaper texts were taken from Herald Glasgow, with permission from Herald & Times Group. Each speaker has a different set of the newspaper texts selected based a greedy algorithm that increases the contextual and phonetic coverage. The details of the text selection algorithms are described in the following paper: C. Veaux, J. Yamagishi and S. King, "The voice bank corpus: Design, collection and data analysis of a large regional accent speech database," https://doi.org/10.1109/ICSDA.2013.6709856. The rainbow passage and elicitation paragraph are the same for all speakers. The rainbow passage can be found at International Dialects of English Archive: (http://web.ku.edu/~idea/readings/rainbow.htm). The elicitation paragraph

326 PAPERS • 5 BENCHMARKS

Speech Commands is an audio dataset of spoken words designed to help train and evaluate keyword spotting systems .

295 PAPERS • 4 BENCHMARKS

Common Voice is an audio dataset that consists of a unique MP3 and corresponding text file. There are 9,283 recorded hours in the dataset. The dataset also includes demographic metadata like age, sex, and accent. The dataset consists of 7,335 validated hours in 60 languages.

267 PAPERS • 264 BENCHMARKS

The ESC-50 dataset is a labeled collection of 2000 environmental audio recordings suitable for benchmarking methods of environmental sound classification. It comprises 2000 5s-clips of 50 different classes across natural, human and domestic sounds, again, drawn from Freesound.org.

246 PAPERS • 6 BENCHMARKS

The ATIS (Airline Travel Information Systems) is a dataset consisting of audio recordings and corresponding manual transcripts about humans asking for flight information on automated airline travel inquiry systems. The data consists of 17 unique intent categories. The original split contains 4478, 500 and 893 intent-labeled reference utterances in train, development and test set respectively.

240 PAPERS • 7 BENCHMARKS

This is a public domain speech dataset consisting of 13,100 short audio clips of a single speaker reading passages from 7 non-fiction books. A transcription is provided for each clip. Clips vary in length from 1 to 10 seconds and have a total length of approximately 24 hours. The texts were published between 1884 and 1964, and are in the public domain. The audio was recorded in 2016-17 by the LibriVox project and is also in the public domain.

233 PAPERS • 3 BENCHMARKS

The SNIPS Natural Language Understanding benchmark is a dataset of over 16,000 crowdsourced queries distributed among 7 user intents of various complexity:

220 PAPERS • 6 BENCHMARKS

MuST-C currently represents the largest publicly available multilingual corpus (one-to-many) for speech translation. It covers eight language directions, from English to German, Spanish, French, Italian, Dutch, Portuguese, Romanian and Russian. The corpus consists of audio, transcriptions and translations of English TED talks, and it comes with a predefined training, validation and test split.

180 PAPERS • 2 BENCHMARKS

The Lip Reading in the Wild (LRW) dataset a large-scale audio-visual database that contains 500 different words from over 1,000 speakers. Each utterance has 29 frames, whose boundary is centered around the target word. The database is divided into training, validation and test sets. The training set contains at least 800 utterances for each class while the validation and test sets contain 50 utterances.

165 PAPERS • 7 BENCHMARKS

LibriTTS is a multi-speaker English corpus of approximately 585 hours of read English speech at 24kHz sampling rate, prepared by Heiga Zen with the assistance of Google Speech and Google Brain team members. The LibriTTS corpus is designed for TTS research. It is derived from the original materials (mp3 audio files from LibriVox and text files from Project Gutenberg) of the LibriSpeech corpus. The main differences from the LibriSpeech corpus are listed below:

151 PAPERS • 1 BENCHMARK

MUSAN is a corpus of music, speech and noise. This dataset is suitable for training models for voice activity detection (VAD) and music/speech discrimination. The dataset consists of music from several genres, speech from twelve languages, and a wide assortment of technical and non-technical noises.

151 PAPERS • NO BENCHMARKS YET

AudioCaps is a dataset of sounds with event descriptions that was introduced for the task of audio captioning, with sounds sourced from the AudioSet dataset. Annotators were provided the audio tracks together with category hints (and with additional video hints if needed).

116 PAPERS • 6 BENCHMARKS

The SumMe dataset is a video summarization dataset consisting of 25 videos, each annotated with at least 15 human summaries (390 in total).

115 PAPERS • 3 BENCHMARKS

Urban Sound 8K is an audio dataset that contains 8732 labeled sound excerpts (<=4s) of urban sounds from 10 classes: air_conditioner, car_horn, children_playing, dog_bark, drilling, enginge_idling, gun_shot, jackhammer, siren, and street_music. The classes are drawn from the urban sound taxonomy. All excerpts are taken from field recordings uploaded to www.freesound.org.

113 PAPERS • 1 BENCHMARK

Consists of more than 210k videos for 310 audio classes.

113 PAPERS • 2 BENCHMARKS

NSynth is a dataset of one shot instrumental notes, containing 305,979 musical notes with unique pitch, timbre and envelope. The sounds were collected from 1006 instruments from commercial sample libraries and are annotated based on their source (acoustic, electronic or synthetic), instrument family and sonic qualities. The instrument families used in the annotation are bass, brass, flute, guitar, keyboard, mallet, organ, reed, string, synth lead and vocal. Four second monophonic 16kHz audio snippets were generated (notes) for the instruments.

106 PAPERS • 1 BENCHMARK

Clotho is an audio captioning dataset, consisting of 4981 audio samples, and each audio sample has five captions (a total of 24 905 captions). Audio samples are of 15 to 30 s duration and captions are eight to 20 words long.

105 PAPERS • 3 BENCHMARKS

Freesound Dataset 50k (or FSD50K for short) is an open dataset of human-labeled sound events containing 51,197 Freesound clips unequally distributed in 200 classes drawn from the AudioSet Ontology. FSD50K has been created at the Music Technology Group of Universitat Pompeu Fabra. It consists mainly of sound events produced by physical sound sources and production mechanisms, including human sounds, sounds of things, animals, natural sounds, musical instruments and more.

100 PAPERS • 2 BENCHMARKS

This dataset contains 118,081 short video clips extracted from 202 movies. Each video has a caption, either extracted from the movie script or from transcribed DVS (descriptive video services) for the visually impaired. The validation set contains 7408 clips and evaluation is performed on a test set of 1000 videos from movies disjoint from the training and val sets.

98 PAPERS • 4 BENCHMARKS

The Oxford-BBC Lip Reading Sentences 2 (LRS2) dataset is one of the largest publicly available datasets for lip reading sentences in-the-wild. The database consists of mainly news and talk shows from BBC programs. Each sentence is up to 100 characters in length. The training, validation and test sets are divided according to broadcast date. It is a challenging set since it contains thousands of speakers without speaker labels and large variation in head pose. The pre-training set contains 96,318 utterances, the training set contains 45,839 utterances, the validation set contains 1,082 utterances and the test set contains 1,242 utterances.

87 PAPERS • 9 BENCHMARKS

The MUSDB18 is a dataset of 150 full lengths music tracks (~10h duration) of different genres along with their isolated drums, bass, vocals and others stems.

84 PAPERS • 1 BENCHMARK

The Free Music Archive (FMA) is a large-scale dataset for evaluating several tasks in Music Information Retrieval. It consists of 343 days of audio from 106,574 tracks from 16,341 artists and 14,854 albums, arranged in a hierarchical taxonomy of 161 genres. It provides full-length and high-quality audio, pre-computed features, together with track- and user-level metadata, tags, and free-form text such as biographies.

82 PAPERS • 1 BENCHMARK

The MAESTRO dataset contains over 200 hours of paired audio and MIDI recordings from ten years of International Piano-e-Competition. The MIDI data includes key strike velocities and sustain/sostenuto/una corda pedal positions. Audio and MIDI files are aligned with ∼3 ms accuracy and sliced to individual musical pieces, which are annotated with composer, title, and year of performance. Uncompressed audio is of CD quality or higher (44.1–48 kHz 16-bit PCM stereo).

76 PAPERS • NO BENCHMARKS YET

The sleep-edf database contains 197 whole-night PolySomnoGraphic sleep recordings, containing EEG, EOG, chin EMG, and event markers. Some records also contain respiration and body temperature. Corresponding hypnograms (sleep patterns) were manually scored by well-trained technicians according to the Rechtschaffen and Kales manual, and are also available.

71 PAPERS • 5 BENCHMARKS

VQG is a collection of datasets for visual question generation. VQG questions were collected by crowdsourcing the task on Amazon Mechanical Turk (AMT). The authors provided details on the prompt and the specific instructions for all the crowdsourcing tasks in this paper in the supplementary material. The prompt was successful at capturing nonliteral questions. Images were taken from the MSCOCO dataset.

68 PAPERS • 1 BENCHMARK

The How2 dataset contains 13,500 videos, or 300 hours of speech, and is split into 185,187 training, 2022 development (dev), and 2361 test utterances. It has subtitles in English and crowdsourced Portuguese translations.

66 PAPERS • 2 BENCHMARKS

The MetaQA dataset consists of a movie ontology derived from the WikiMovies Dataset and three sets of question-answer pairs written in natural language: 1-hop, 2-hop, and 3-hop queries.

53 PAPERS • 1 BENCHMARK

We present the MSP-IMPROV corpus, a multimodal emotional database, where the goal is to have control over lexical content and emotion while also promoting naturalness in the recordings. Studies on emotion perception often require stimuli with fixed lexical content, but that convey different emotions. These stimuli can also serve as an instrument to understand how emotion modulates speech at the phoneme level, in a manner that controls for coarticulation. Such audiovisual data are not easily available from natural recordings. A common solution is to record actors reading sentences that portray different emotions, which may not produce natural behaviors. We propose an alternative approach in which we define hypothetical scenarios for each sentence that are carefully designed to elicit a particular emotion. Two actors improvise these emotion-specific situations, leading them to utter contextualized, non-read renditions of sentences that have fixed lexical content and convey different emot

52 PAPERS • 1 BENCHMARK

GigaSpeech, an evolving, multi-domain English speech recognition corpus with 10,000 hours of high quality labeled audio suitable for supervised training, and 40,000 hours of total audio suitable for semi-supervised and unsupervised training.

50 PAPERS • 3 BENCHMARKS

TED-LIUM 3 is an audio dataset collected from TED Talks. It contains:

50 PAPERS • NO BENCHMARKS YET

The SEMAINE videos dataset contains spontaneous data capturing the audiovisual interaction between a human and an operator undertaking the role of an avatar with four personalities: Poppy (happy), Obadiah (gloomy), Spike (angry) and Prudence (pragmatic). The audiovisual sequences have been recorded at a video rate of 25 fps (352 x 288 pixels). The dataset consists of audiovisual interaction between a human and an operator undertaking the role of an agent (Sensitive Artificial Agent). SEMAINE video clips have been annotated with couples of epistemic states such as agreement, interested, certain, concentration, and thoughtful with continuous rating (within the range [1,-1]) where -1 indicates most negative rating (i.e: No concentration at all) and +1 defines the highest (Most concentration). Twenty-four recording sessions are used in the Solid SAL scenario. Recordings are made of both the user and the operator, and there are usually four character interactions in each recording session,

48 PAPERS • 1 BENCHMARK

Fluent Speech Commands is an open source audio dataset for spoken language understanding (SLU) experiments. Each utterance is labeled with "action", "object", and "location" values; for example, "turn the lights on in the kitchen" has the label {"action": "activate", "object": "lights", "location": "kitchen"}. A model must predict each of these values, and a prediction for an utterance is deemed to be correct only if all values are correct.

44 PAPERS • 1 BENCHMARK

The REVERB (REverberant Voice Enhancement and Recognition Benchmark) challenge is a benchmark for evaluation of automatic speech recognition techniques. The challenge assumes the scenario of capturing utterances spoken by a single stationary distant-talking speaker with 1-channe, 2-channel or 8-channel microphone-arrays in reverberant meeting rooms. It features both real recordings and simulated data.

44 PAPERS • 1 BENCHMARK

The VOICES corpus is a dataset to promote speech and signal processing research of speech recorded by far-field microphones in noisy room conditions.

43 PAPERS • NO BENCHMARKS YET

Multilingual LibriSpeech is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish. It includes about 44.5K hours of English and a total of about 6K hours for other languages.

40 PAPERS • 7 BENCHMARKS

MagnaTagATune dataset contains 25,863 music clips. Each clip is a 29-seconds-long excerpt belonging to one of the 5223 songs, 445 albums and 230 artists. The clips span a broad range of genres like Classical, New Age, Electronica, Rock, Pop, World, Jazz, Blues, Metal, Punk, and more. Each audio clip is supplied with a vector of binary annotations of 188 tags. These annotations are obtained by humans playing the two-player online TagATune game. In this game, the two players are either presented with the same or a different audio clip. Subsequently, they are asked to come up with tags for their specific audio clip. Afterward, players view each other’s tags and are asked to decide whether they were presented the same audio clip. Tags are only assigned when more than two players agreed. The annotations include tags like ’singer’, ’no singer’, ’violin’, ’drums’, ’classical’, ’jazz’. The top 50 most popular tags are typically used for evaluation to ensure that there is enough training data f

39 PAPERS • 2 BENCHMARKS

The RWC (Real World Computing) Music Database is a copyright-cleared music database (DB) that is available to researchers as a common foundation for research. It contains around 100 complete songs with manually labeled section boundaries. For the 50 instruments, individual sounds at half-tone intervals were captured with several variations of playing styles, dynamics, instrument manufacturers and musicians.

39 PAPERS • NO BENCHMARKS YET

We propose Localized Narratives, a new form of multimodal image annotations connecting vision and language. We ask annotators to describe an image with their voice while simultaneously hovering their mouse over the region they are describing. Since the voice and the mouse pointer are synchronized, we can localize every single word in the description. This dense visual grounding takes the form of a mouse trace segment per word and is unique to our data. We annotated 849k images with Localized Narratives: the whole COCO, Flickr30k, and ADE20K datasets, and 671k images of Open Images, all of which we make publicly available. We provide an extensive analysis of these annotations showing they are diverse, accurate, and efficient to produce. We also demonstrate their utility on the application of controlled image captioning.

38 PAPERS • 5 BENCHMARKS

MedleyDB, is a dataset of annotated, royalty-free multitrack recordings. It was curated primarily to support research on melody extraction. For each song melody f₀ annotations are provided as well as instrument activations for evaluating automatic instrument recognition. The original dataset consists of 122 multitrack songs out of which 108 include melody annotations.

37 PAPERS • NO BENCHMARKS YET

WHAMR! is a dataset for noisy and reverberant speech separation. It extends WHAM! by introducing synthetic reverberation to the speech sources in addition to the existing noise. Room impulse responses were generated and convolved using pyroomacoustics. Reverberation times were chosen to approximate domestic and classroom environments (expected to be similar to the restaurants and coffee shops where the WHAM! noise was collected), and further classified as high, medium, and low reverberation based on a qualitative assessment of the mixture’s noise recording.

36 PAPERS • 3 BENCHMARKS

We introduce FLEURS, the Few-shot Learning Evaluation of Universal Representations of Speech benchmark. FLEURS is an n-way parallel speech dataset in 102 languages built on top of the machine translation FLoRes-101 benchmark, with approximately 12 hours of speech supervision per language. FLEURS can be used for a variety of speech tasks, including Automatic Speech Recognition (ASR), Speech Language Identification (Speech LangID), Translation and Retrieval. In this paper, we provide baselines for the tasks based on multilingual pre-trained models like mSLAM. The goal of FLEURS is to enable speech technology in more languages and catalyze research in low-resource speech understanding.

33 PAPERS • 2 BENCHMARKS

The UCR Time Series Archive - introduced in 2002, has become an important resource in the time series data mining community, with at least one thousand published papers making use of at least one data set from the archive. The original incarnation of the archive had sixteen data sets but since that time, it has gone through periodic expansions. The last expansion took place in the summer of 2015 when the archive grew from 45 to 85 data sets. This paper introduces and will focus on the new data expansion from 85 to 128 data sets. Beyond expanding this valuable resource, this paper offers pragmatic advice to anyone who may wish to evaluate a new algorithm on the archive. Finally, this paper makes a novel and yet actionable claim: of the hundreds of papers that show an improvement over the standard baseline (1-nearest neighbor classification), a large fraction may be misattributing the reasons for their improvement. Moreover, they may have been able to achieve the same improvement with a

32 PAPERS • 2 BENCHMARKS