Self-supervised learning (SSL) has helped extend speech technologies to more languages by reducing the need for labeled data. However, models are still far from supporting the world’s 7000+ languages. We propose XEUS, a Cross-lingual Encoder for Universal Speech, trained on over 1 million hours of data across 4057 languages, extending the language coverage of SSL models 4-fold. We combine 1 million hours of speech from existing publicly accessible corpora with a newly created corpus of 7400+ hours from 4057 languages, which will be publicly released. To handle the diverse conditions of multilingual speech data, we augment the typical SSL masked prediction approach with a novel dereverberation objective, increasing robustness. We evaluate XEUS on several benchmarks, and show that it consistently outperforms or achieves comparable results to state-of-the-art (SOTA) SSL models across a variety of tasks. XEUS sets a new SOTA on the ML-SUPERB benchmark: it outperforms MMS 1B and w2v-BERT 2.0 v2 by 0.8% and 4.4% respectively, despite having less parameters or pre-training data. Checkpoints, code, and data are found in https://www.wavlab.org/activities/2024/xeus/.
This work describes CMU’s submission to the IWSLT 2024 Offline Speech Translation (ST) Shared Task for translating English speech to German, Chinese, and Japanese text. We are the first participants to employ a long-form strategy which directly processes unsegmented recordings without the need for a separate voice-activity detection stage (VAD). We show that the Whisper automatic speech recognition (ASR) model has a hallucination problem when applied out-of-the-box to recordings containing non-speech noises, but a simple noisy fine-tuning approach can greatly enhance Whisper’s long-form robustness across multiple domains. Then, we feed English ASR outputs into fine-tuned NLLB machine translation (MT) models which are decoded using COMET-based Minimum Bayes Risk. Our VAD-free ASR+MT cascade is tested on TED talks, TV series, and workout videos and shown to outperform prior winning IWSLT submissions and large open-source models.
Large language models (LLMs) have successfully served as a general-purpose interface across multiple tasks and languages, while the adaptation of voice LLMs is mostly designed for specific purposes (either single-task or monolingual), where the advantages of LLMs especially for low-resource language processing and zero-shot task generalization are less exploited in the audio community. To bridge the gap, we introduce Make-A-Voice as a multi-modal voice LLM and conduct a comprehensive study on its capability to deal with multiple tasks/languages. When trained on ~200K hours of 6-language data for 4 voice generation applications, Make-A-Voice emerges notable advantages: 1) as scalable learners to improve performance with end-to-end local and global multiscale transformers; and 2) as multitask learners by adjusting prompts to share common knowledge across modalities (speech/singing) and present in-context learning abilities by generalizing to unseen tasks not explicitly train on; 3) as multilingual learners to alleviate data scarcity of low-resource languages by including rich-resource language training data. Experimental results demonstrate that Make-A-Voice exhibits superior audio quality and style similarity compared with competitive baseline models in monolingual/cross-lingual voice generation. Audio samples are available at https://M-Voice.github.io
This paper presents the extscMineTrans English-to-Chinese speech translation systems developed for two challenge tracks of IWSLT 2023, i.e., Offline Speech Translation (S2T) and Speech-to-Speech Translation (S2ST). For the S2T track, extscMineTrans employs a practical cascaded system to explore the limits of translation performance in both constrained and unconstrained settings, where the whole system consists of automatic speech recognition (ASR), punctuation recognition (PC), and machine translation (MT) modules. We also investigate the effectiveness of multiple ASR architectures and explore two MT strategies: supervised in-domain fine-tuning and prompt-guided translation using a large language model. For the S2ST track, we explore a speech-to-unit (S2U) framework to build an end-to-end S2ST system. This system encodes the target speech as discrete units via our trained HuBERT. Then it leverages the standard sequence-to-sequence model to directly learn the mapping between source speech and discrete units without any auxiliary recognition tasks (i.e., ASR and MT tasks). Various efforts are made to improve the extscMineTrans’s performance, such as acoustic model pre-training on large-scale data, data filtering, data augmentation, speech segmentation, knowledge distillation, consistency training, model ensembles, etc.