Existing large-scale zero-shot text-to-speech (TTS) models deliver high speech quality but suffer from slow inference speeds due to massive parameters. To address this issue, this paper introduces ZipVoice, a high-quality flow-matching-based zero-shot TTS model with a compact model size and fast inference speed. Key designs include: 1) a Zipformer-based vector field estimator to maintain adequate modeling capabilities under constrained size; 2) Average upsampling-based initial speech-text alignment and Zipformer-based text encoder to improve speech intelligibility; 3) A flow distillation method to reduce sampling steps and eliminate the inference overhead associated with classifier-free guidance. Experiments on 100 k hours multilingual datasets show that ZipVoice matches state-of-the-art models in speech quality, while being 3 times smaller and up to 30 times faster than a DiT-based flow-matching baseline. Codes, model checkpoints and demo samples are publicly available.11https://github.com/k2-fsa/ZipVoice

Connectionist Temporal Classification (CTC) is a widely used method for automatic speech recognition (ASR), renowned for its simplicity and computational efficiency. However, it often falls short in recognition performance. In this work, we propose the Consistency-Regularized CTC (CR-CTC), which enforces consistency between two CTC distributions obtained from different augmented views of the input speech mel-spectrogram. We provide in-depth insights into its essential behaviors from three perspectives: 1) it conducts self-distillation between random pairs of sub-models that process different augmented views; 2) it learns contextual representation through masked prediction for positions within time-masked regions, especially when we increase the amount of time masking; 3) it suppresses the extremely peaky CTC distributions, thereby reducing overfitting and improving the generalization ability. Extensive experiments on LibriSpeech, Aishell-1, and GigaSpeech datasets demonstrate the effectiveness of our CR-CTC. It significantly improves the CTC performance, achieving state-of-the-art results comparable to those attained by transducer or systems combining CTC and attention-based encoder-decoder (CTC/AED). We release our code at \url{https://github.com/k2-fsa/icefall}.

Self-supervised learning (SSL) has achieved great success in speech-related tasks. While Transformer and Conformer architectures have dominated SSL backbones, encoders like Zipformer, which excel in automatic speech recognition (ASR), remain unexplored in SSL. Concurrently, inefficiencies in data processing within existing SSL training frameworks, such as fairseq, pose challenges in managing the growing volumes of training data. To address these issues, we propose k2SSL, an open-source framework that offers faster, more memory-efficient, and better-performing self-supervised speech representation learning, focusing on downstream ASR tasks. The optimized HuBERT and proposed Zipformer-based SSL systems exhibit substantial reductions in both training time and memory usage during SSL training. Experiments on LibriSpeech demonstrate that Zipformer Base significantly outperforms HuBERT and WavLM, achieving up to a 34.8% relative WER reduction compared to HuBERT Base after fine-tuning, along with a 3.5x pre-training speedup in GPU hours. When scaled to 60k hours of LibriLight data, Zipformer Large exhibits remarkable efficiency, matching HuBERT Large’s performance while requiring only 5/8 pre-training steps.

Modern large-language models often face communication bottlenecks on current hardware rather than computational limitations. *Multi-head latent attention (MLA)* addresses this by compressing the key-value cache using low-rank matrices, while the Absorb operation prevents the KV cache from reverting to its original size, significantly boosting both training and inference speed. Despite the success of DeepSeek V2/V3/R1, most model providers have heavily invested in optimizing GQA-based models and, therefore, lack strong incentives to retrain MLA-based models from scratch. This paper demonstrates that MLA provides superior expressive power compared to GQA with the same KV cache overhead, thereby offering a rationale for transitioning from GQA to MLA. In addition, we introduce TransMLA, a framework that seamlessly converts any GQA-based pre-trained model (e.g., LLaMA, Qwen, Gemma, Mistral/Mixtral) into an MLA-based model. For the first time, our method enables *direct conversion of these models into a format compatible with DeepSeek's codebase*, allowing them to fully leverage the existing, highly-optimized support for the DeepSeek architecture within inference engines like vLLM and SGlang. By compressing 93\% of the KV cache in LLaMA-2-7B, we achieve a **10x speedup** with an 8K context length while maintaining meaningful output. Moreover, the model requires only **6B tokens** for fine-tuning to recover comparable performance across multiple benchmarks. TransMLA provides a practical path for migrating GQA-based models to the MLA structure, and when combined with DeepSeek’s advanced optimizations—such as FP8 quantization and Multi-Token Prediction—further inference acceleration can be achieved.

In this paper, we introduce Libriheavy, a large-scale ASR corpus consisting of 50,000 hours of read English speech derived from LibriVox. To the best of our knowledge, Libriheavy is the largest freely-available corpus of speech with supervisions. Different from other open-sourced datasets that only provide normalized transcriptions, Libriheavy contains richer information such as punctuation, casing and text context, which brings more flexibility for system building. Specifically, we propose a general and efficient pipeline to locate, align and segment the audios in previously published Librilight to its corresponding texts. The same as Librilight, Libriheavy also has three training subsets small, medium, large of the sizes 500h, 5000h, 50000h respectively. We also extract the dev and test evaluation sets from the aligned audios and guarantee there is no overlapping speakers and books in training sets. Baseline systems are built on the popular CTC-Attention and transducer models. Additionally, we open-source our dataset creatation pipeline which can also be used to other audio alignment tasks.

Prompts are crucial to large language models as they provide context information such as topic or logical relationships. Inspired by this, we propose PromptASR, a framework that integrates prompts in end-to-end automatic speech recognition (E2E ASR) systems to achieve contextualized ASR with controllable style of transcriptions. Specifically, a dedicated text encoder encodes the text prompts and the encodings are injected into the speech encoder by cross-attending the features from two modalities. When using the ground truth text from preceding utterances as content prompt, the proposed system achieves 21.9% and 6.8% relative word error rate reductions on a book reading dataset and an in-house dataset compared to a baseline ASR system. The system can also take word-level biasing lists as prompt to improve recognition accuracy on rare words. An additional style prompt can be given to the text encoder and guide the ASR system to output different styles of transcriptions. The code is available at icefall1.

The Conformer has become the most popular encoder model for automatic speech recognition (ASR). It adds convolution modules to a transformer to learn both local and global dependencies. In this work we describe a faster, more memory-efficient, and better-performing transformer, called Zipformer. Modeling changes include: 1) a U-Net-like encoder structure where middle stacks operate at lower frame rates; 2) reorganized block structure with more modules, within which we re-use attention weights for efficiency; 3) a modified form of LayerNorm called BiasNorm allows us to retain some length information; 4) new activation functions SwooshR and SwooshL work better than Swish. We also propose a new optimizer, called ScaledAdam, which scales the update by each tensor's current scale to keep the relative change about the same, and also explictly learns the parameter scale. It achieves faster converge and better performance than Adam. Extensive experiments on LibriSpeech, Aishell-1, and WenetSpeech datasets demonstrate the effectiveness of our proposed Zipformer over other state-of-the-art ASR models. Our code is publicly available at https://github.com/k2-fsa/icefall.

Knowledge distillation (KD) is a common approach to improve model performance in automatic speech recognition (ASR), where a student model is trained to imitate the output behaviour of a teacher model. However, traditional KD methods suffer from teacher label storage issue, especially when the training corpora are large. Although on-the-fly teacher label generation tackles this issue, the training speed is significantly slower as the teacher model has to be evaluated every batch. In this paper, we reformulate the generation of teacher label as a codec problem. We propose a novel Multi-codebook Vector Quantization (MVQ) approach that compresses teacher embeddings to codebook indexes (CI). Based on this, a KD training framework (MVQ-KD) is proposed where a student model predicts the CI generated from the embeddings of a self-supervised pre-trained teacher model. Experiments on the LibriSpeech clean-100 hour show that MVQ-KD framework achieves comparable performance as traditional KD methods (11, 12), while requiring 256 times less storage. When the full LibriSpeech dataset is used, MVQ-KD framework results in 13.8% and 8.2% relative word error rate reductions (WERRs) for non -streaming transducer on test-clean and test-other and 4.0% and 4.9% for streaming transducer. The implementation of this work is already released as a part of the open-source project icefall1.

The transducer architecture is becoming increasingly popular in the field of speech recognition, because it is naturally streaming as well as high in accuracy. One of the drawbacks of transducer is that it is difficult to decode in a fast and parallel way due to an unconstrained number of symbols that can be emitted per time step.In this work, we introduce a constrained version of transducer loss to learn strictly monotonic alignments between the sequences; we also improve the standard greedy search and beam search algorithms by limiting the number of symbols that can be emitted per time step in transducer decoding, making it more efficient to decode in parallel with batches. Furthermore, we propose an finite state automaton-based (FSA) parallel beam search algorithm that can run with graphs on GPU efficiently. The experiment results show that we achieve slight word error rate (WER) improvement as well as significant speedup in decoding. Our work is open-sourced and publicly available1.

In streaming automatic speech recognition (ASR), it is desirable to reduce latency as much as possible while having minimum impact on recognition accuracy. Although a few existing methods are able to achieve this goal, they are difficult to implement due to their dependency on external alignments. In this paper, we propose a simple way to penalize symbol delay in transducer model, so that we can balance the trade-off between symbol delay and accuracy for streaming models without external alignments. Specifically, our method adds a small constant times (T/2 - t), where T is the number of frames and t is the current frame, to all the non-blank log-probabilities (after normalization) that are fed into the two dimensional transducer recursion. For both streaming Conformer models and unidirectional long short-term memory (LSTM) models, experimental results show that it can significantly reduce the symbol delay with an acceptable performance degradation. Our method achieves similar delay-accuracy trade-off to the previously published FastEmit, but we believe our method is preferable because it has a better justification: it is equivalent to penalizing the average symbol delay. Our work is open-sourced and publicly available1.

The RNN-Transducer (RNN-T) framework for speech recognition has been growing in popularity, particularly for deployed real-time ASR systems, because it combines high accuracy with naturally streaming recognition.One of the drawbacks of RNN-T is that its loss function is relatively slow to compute, and can use a lot of memory.Excessive GPU memory usage can make it impractical to use RNN-T loss in cases where the vocabulary size is large: for example, for Chinese character-based ASR.We introduce a method for faster and more memoryefficient RNN-T loss computation.We first obtain pruning bounds for the RNN-T recursion using a simple joiner network that is linear in the encoder and decoder embeddings; we can evaluate this without using much memory.We then use those pruning bounds to evaluate the full, non-linear joiner network.The code is open-sourced and publicly available.

The crux of single-channel speech separation is how to encode the mixture of signals into such a latent embedding space that the signals from different speakers can be precisely separated. Existing methods for speech separation either transform the speech signals into frequency domain to perform separation or seek to learn a separable embedding space by constructing a latent domain based on convolutional filters. While the latter type of methods learning an embedding space achieves substantial improvement for speech separation, we argue that the embedding space defined by only one latent domain does not suffice to provide a thoroughly separable encoding space for speech separation. In this paper, we propose the Stepwise-Refining Speech Separation Network (SRSSN), which follows a coarse-to-fine separation framework. It first learns a 1-order latent domain to define an encoding space and thereby performs a rough separation in the coarse phase. Then the proposedSRSSNlearns a new latent domain along each basis function of the existing latent domain to obtain a high-order latent domain in the refining phase, which enables our model to perform a refining separation to achieve a more precise speech separation. We demonstrate the effectiveness of ourSRSSNby conducting extensive experiments, including speech separation in a clean (noise-free) setting on WSJ0-2/3mix datasets as well as in noisy/reverberant settings on WHAM!/WHAMR! datasets. Furthermore, we also perform experiments of speech recognition on separated speech signals by our model to evaluate the performance of speech separation indirectly.

BACKGROUND: Fingerprint biometrics play an essential role in authentication. It remains a challenge to match fingerprints with the minutiae or ridges missing. Many fingerprints failed to match their targets due to the incompleteness. RESULT: In this work, we modeled the fingerprints with Bezier curves and proposed a novel algorithm to detect and restore fragmented ridges in incomplete fingerprints. In the proposed model, the Bezier curves' control points represent the fingerprint fragments, reducing the data size by 89% compared to image representations. The representation is lossless as the restoration from the control points fully recovering the image. Our algorithm can effectively restore incomplete fingerprints. In the SFinGe synthetic dataset, the fingerprint image matching score increased by an average of 39.54%, the ERR (equal error rate) is 4.59%, and the FMR1000 (false match rate) is 2.83%, these are lower than 6.56% (ERR) and 5.93% (FMR1000) before restoration. In FVC2004 DB1 real fingerprint dataset, the average matching score increased by 13.22%. The ERR reduced from 8.46% before restoration to 7.23%, and the FMR1000 reduced from 20.58 to 18.01%. Moreover, We assessed the proposed algorithm against FDP-M-net and U-finger in SFinGe synthetic dataset, where FDP-M-net and U-finger are both convolutional neural network models. The results show that the average match score improvement ratio of FDP-M-net is 1.39%, U-finger is 14.62%, both of which are lower than 39.54%, yielded by our algorithm. CONCLUSIONS: Experimental results show that the proposed algorithm can successfully repair and reconstruct ridges in single or multiple damaged regions of incomplete fingerprint images, and hence improve the accuracy of fingerprint matching.

Speech emotion recognition plays an increasingly important role in emotional computing and is still a challenging task due to its complexity. In this study, we developed a framework integrating three distinctive classifiers: a deep neural network (DNN), a convolution neural network (CNN), and a recurrent neural network (RNN). The framework was used for categorical recognition of four discrete emotions (i.e., angry, happy, neutral and sad). Frame-level low-level descriptors (LLDs), segment-level mel-spectrograms (MS), and utterance-level outputs of high-level statistical functions (HSFs) on LLDs were passed to RNN, CNN, and DNN, separately. Three individual models of LLD-RNN, MS-CNN, and HSF-DNN were obtained. In the models of MS-CNN and LLD-RNN, the attention mechanism based weighted-pooling method was utilized to aggregate the CNN and RNN outputs. To effectively utilize the interdependencies between the two approaches of emotion description (discrete emotion categories and continuous emotion attributes), a multi-task learning strategy was implemented in these three models to acquire generalized features by simultaneously operating classification of discrete categories and regression of continuous attributes. Finally, a confidence-based fusion strategy was developed to integrate the power of different classifiers in recognizing different emotional states. Three experiments on emotion recognition based on the IEMOCAP corpus were conducted. Our experimental results show that the weighted pooling method based on attention mechanism endowed the neural networks with the capability to focus on emotionally salient parts. The generalized features learned in the multi-task learning helped the neural networks to achieve higher accuracies in the tasks of emotion classification. Furthermore, our proposed fusion system achieved weighted accuracy of 57.1% and unweighted accuracy of 58.3%, which were significantly higher than those of each individual classifier. The effectiveness of the proposed approach based on classifier fusion was thus validated.