FLEURS: FEW-Shot Learning Evaluation of Universal Representations of Speech

By Alexis Conneau, Min Ma, Simran Khanuja, Yu Zhang, Vera Axelrod, Siddharth Dalmia, Jason Riesa, Clara Rivera, Ankur Bapna. 2022

651 citations recorded by the source corpus.

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), Speech-Text Retrieval. In this paper, we provide baselines for the tasks based on multilingual pre-trained models like speech-only w2v-BERT [1] and speech-text multimodal mSLAM [2]. The goal of FLEURS is to enable speech technology in more languages and catalyze research in low-resource speech understanding.1.

Subfields: Artificial Intelligence, Evaluation & Benchmarks, Speech & Audio

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yu Zhang
2. Jinyu Li
3. Wei-Ning Hsu
4. Tara N. Sainath
5. Graham Neubig
6. Quoc V. Le
7. Yonghui Wu
8. Dong Yu

In Evaluation & Benchmarks

1. Sebastian Ruder
2. Melvin Johnson
3. Yu Zhang
4. Alexis Conneau
5. Orhan Firat
6. Alexei Baevski
7. Colin Cherry
8. Holger Schwenk

In Speech & Audio

1. Jinyu Li
2. Yu Zhang
3. Wei-Ning Hsu
4. Tara N. Sainath
5. Yonghui Wu
6. Chung-Cheng Chiu
7. Alexei Baevski
8. Ye Jia

Citations in the corpus

• Robust Speech Recognition via Large-Scale Weak Supervision (2022, methodology): “To study the performance of Whisper more broadly we also report performance on the Fleurs dataset (Conneau et al., 2022).”
• AudioPaLM: A Large Language Model That Can Speak and Listen (2023, positive): “Table 17: Zero-shot AST performance on FLEURS.”
• Speech Recognition via Large-Scale Weak Supervision (neutral): “The open OWSM system is trained on 180k hours, mainly hand-transcribed publicly available data, including such datasets as LibriSpeech and Multilingual LibriSpeech, Common Voice, FLEURS, Switchboard, AMI, and others; see (Peng et al., 2023) for details.”
• Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs (2025, neutral): “We evaluate the AST performance on two public benchmarks: CoVoST2 [WWGP21] and FLEURS [CMK + 23].”
• Scaling Speech Technology to 1, 000+ Languages (2023, methodology): “We compare to BABEL [Gales et al., 2014], CMU Wilderness [Black, 2019], CommonVoice [Ardila et al., 2020], Dhwani [Javed et al., 2022], FLEURS [Conneau et al., 2022], M-AILABS [M-AILABS, 2018], MLS [Pratap et al., 2020c], VoxLingua107 [Valk and Alumäe, 2020], and VoxPopuli [Wang et al., 2021] .…”
• Google USM: Scaling Automatic Speech Recognition Beyond 100 Languages (2023, methodology): “The FLEURS [16] dataset is a publicly available, multi-way parallel dataset of 10 hours of read speech in 102 languages spanning 7 geo-groups.”
• SpeechMatrix: A Large-Scale Mined Corpus of Multilingual Speech-to-Speech Translations (2022, background): “(3) FLEURS (Conneau et al., 2022).”
• XTREME-S: Evaluating Cross-lingual Speech Representations (2022, background): “Languages and results are also grouped into seven geographical areas: Western Europe (WE), Eastern Europe (EE), Central-Asian/Middle-East/North-Africa (CMN), Sub-Saharan Africa (SSA), South Asia (SA), South-Eastern Asia (SEA) and CJK languages (CJK) [37].”