Search

The spin Seebeck effect (SSE) is a phenomenon of thermoelectric generation that occurs within a device consisting of a bilayer of a metal and a ferromagnet. When Tb$_3$Fe$_5$O$_{12}$ (TbIG) is substituted for the ferromagnet, the effect goes to zero at low temperatures, yet it increases to positive values with the application of a magnetic field. This is opposite to the expectation that the SSE should be suppressed by a magnetic field due to the increase in the magnon gap. In this paper, the crystal-field excitations (CFE) in TbIG are calculated within a mean field theory exploiting the parameters of Terbium Gallium Garnet Tb$_3$Ga$_5$O$_{12}$ (TGG) from the neutron-scattering experimental literature. Like TGG, the primitive cell of TbIG hosts twelve Tb sites with six inequivalent magnetic sublattices, but due to the net $[111]$-molecular field from the tetrahedral and octahedral Fe ions, these can be classified into two distinct groups, the $\mathbf{C}$ and the $\mathbf{C'}$ sites, which account for the `double umbrella' magnetic structure. We show that when an external magnetic field is applied along the [111] direction of the crystal, the lowest CFE of the $\mathbf{C}$ sublattices decreases. As a consequence of the magnetic field dependence of the lowest CFE, we find that at low temperatures the SSE in TbIG can result enhanced by an applied magnetic field., Comment: 12 pages, 16 figures

Fusing speech into pre-trained language model (SpeechLM) usually suffers from inefficient encoding of long-form speech and catastrophic forgetting of pre-trained text modality. We propose SSR-Connector (Segmented Speech Representation Connector) for better modality fusion. Leveraging speech-text alignments, our approach segments and compresses speech features to match the granularity of text embeddings. Additionally, we introduce a two-stage training pipeline that includes the distillation and fine-tuning phases to mitigate catastrophic forgetting. SSR-Connector outperforms existing mechanism for speech-text modality fusion, consistently achieving better speech understanding (e.g., +10 accuracy on StoryCloze and +20 on Speech-MMLU) while preserving pre-trained text ability.

The electrically readable complex dynamics of robust and scalable magnetic tunnel junctions (MTJs) offer promising opportunities for advancing neuromorphic computing. In this work, we present an MTJ design with a free layer and two polarizers capable of computing the sigmoidal activation function and its gradient at the device level. This design enables both feedforward and backpropagation computations within a single device, extending neuromorphic computing frameworks previously explored in the literature by introducing the ability to perform backpropagation directly in hardware. Our algorithm implementation reveals two key findings: (i) the small discrepancies between the MTJ-generated curves and the exact software-generated curves have a negligible impact on the performance of the backpropagation algorithm, (ii) the device implementation is highly robust to inter-device variation and noise, and (iii) the proposed method effectively supports transfer learning and knowledge distillation. To demonstrate this, we evaluated the performance of an edge computing network using weights from a software-trained model implemented with our MTJ design. The results show a minimal loss of accuracy of only 0.1% for the Fashion MNIST dataset and 2% for the CIFAR-100 dataset compared to the original software implementation. These results highlight the potential of our MTJ design for compact, hardware-based neural networks in edge computing applications, particularly for transfer learning., Comment: 18 pages, 5 figures

In this work, we explored theoretically the spatial resolution of magnetic solitons and the variations of their sizes when subjected to a Magnetic Force Microscopy (MFM) measurement. Next to tip-sample separation, we considered reversal in the magnetization direction of the tip, showing that the magnetic soliton size measurement can be strongly affected by the magnetization direction of the tip. In addition to previous studies that only consider thermal fluctuations, we developed a theoretical method to obtain the minimum observable length of a magnetic soliton and its length variation due to the influence of the MFM tip by minimizing the soliton's magnetic energy. Our model uses analytical and numerical calculations and prevents overestimating the characteristic length scales from MFM images. We compared our method with available data from MFM measurements of domain wall widths, and we performed micromagnetic simulations of a skyrmion-tip system, finding a good agreement for both attractive and repulsive domain wall profile signals and for the skyrmion diameter in the presence of the magnetic tip. Our results provide significant insights for a better interpretation of MFM measurements of different magnetic solitons and will be helpful in the design of potential reading devices based on magnetic solitons as information carriers., Comment: 18 pages. Supplementary Material is included

The classical Ising chain is the paradigm for the non-existence of phase transitions in 1D systems and was solved by Ernst Ising one hundred years ago. More recently, a decorated two leg Ising ladder has received interest for the curious thermodynamics that resemble a phase transition; a sharp peak in the specific heat at low, but finite temperature. We use this model to reveal a bifurcation in the correlation lengths due to a crossing of the sub-leading eigenvalues of the transfer matrix, which results in two distinct length scales necessary to describe to the decay of correlations. We discuss this phenomenon in the context of the geometric frustration in the model. We also provide additional results to aid in the understanding of the curious thermodynamics of the model through a study of the magnetic susceptibilities., Comment: 24 pages, 16 figures. Further references added, misprints corrected, additional points added to conclusions, some ambiguities clarified

Current-driven magnetic skyrmions show promise as carriers of information bits in racetrack magnetic memory applications. Specifically, the utilization of skyrmions in synthetic antiferromagnetic (SAF) systems is highly attractive due to the potential to suppress the Skyrmion Hall effect, which causes a transverse displacement of driven skyrmions relative to the drift direction. In this study, we demonstrate, through analytical calculations and micromagnetic simulations, that in the case of a nanoring geometry, current-driven skyrmions achieve a stable circular motion with a constant frequency, which is a prerequisite for a skyrmion-based clock device. Notably, the operational frequency in a SAF nanoring surpasses that in a bilayer ferromagnetic-heavy metal nanoring and lies in the GHz regime for current densities of 20 MA/cm^2. We also find that the performance of skyrmions in SAF nanorings is comparable to that of radial Neel domain walls for low current densities (approximately 15 MA/cm^2) and low skyrmion densities (Nsk~6). Additionally, we introduce a novel skyrmionic three-phase AC alternator based on a SAF nanoring, which operates at frequencies in the GHz regime. Our findings underscore the potential of SAF nanorings as constituent materials in clock devices with tunable frequencies operating in the GHz regime, Comment: 20 pages, 7 Figures

The growing market and massive use of Internet of Things nodes is placing unprecedented demands of energy efficient hardware for edge computing and microwave devices. In particular, magnetic tunnel junctions (MTJs), as main building blocks of spintronic microwave technology, can offer a path for the development of compact and high-performance microwave detectors. On the other hand, the fascinating field of skyrmionics is bridging together concepts from topology and spintronics. Here, we show the proof of concept of the direct electrical excitation and detection of the dynamics of a topological protected magnetic texture, i.e. skyrmion at room temperature and for a wide region of applied fields, including the zero field case. This topological spin torque diode is realized with an MTJ on top of a skyrmionic material. Quantitative Magnetic Force Microscopy measurements are employed to confirm the existence of a single skyrmion in the MTJ free layer. Spin torque diode electrical measurements show the electrical excitation via spin transfer torque (STT) of a skyrmion resonant mode with frequencies near 4 GHz and a selectivity one order of magnitude smaller than the uniform modes excited in the same device. Micromagnetic simulations identify these dynamics with the excitation of the breathing mode and point out the role of thickness dependent magnetic parameters (magnetic anisotropy field and Dzyaloshinkii Moriya interaction) in both stabilizing and exciting the magnetic skyrmions. This work marks a milestone for the development of topological spin torque diodes.

Large-scale automatic speech translation systems today lack key features that help machine-mediated communication feel seamless when compared to human-to-human dialogue. In this work, we introduce a family of models that enable end-to-end expressive and multilingual translations in a streaming fashion. First, we contribute an improved version of the massively multilingual and multimodal SeamlessM4T model-SeamlessM4T v2. This newer model, incorporating an updated UnitY2 framework, was trained on more low-resource language data. SeamlessM4T v2 provides the foundation on which our next two models are initiated. SeamlessExpressive enables translation that preserves vocal styles and prosody. Compared to previous efforts in expressive speech research, our work addresses certain underexplored aspects of prosody, such as speech rate and pauses, while also preserving the style of one's voice. As for SeamlessStreaming, our model leverages the Efficient Monotonic Multihead Attention mechanism to generate low-latency target translations without waiting for complete source utterances. As the first of its kind, SeamlessStreaming enables simultaneous speech-to-speech/text translation for multiple source and target languages. To ensure that our models can be used safely and responsibly, we implemented the first known red-teaming effort for multimodal machine translation, a system for the detection and mitigation of added toxicity, a systematic evaluation of gender bias, and an inaudible localized watermarking mechanism designed to dampen the impact of deepfakes. Consequently, we bring major components from SeamlessExpressive and SeamlessStreaming together to form Seamless, the first publicly available system that unlocks expressive cross-lingual communication in real-time. The contributions to this work are publicly released and accessible at https://github.com/facebookresearch/seamless_communication

We introduce the Efficient Monotonic Multihead Attention (EMMA), a state-of-the-art simultaneous translation model with numerically-stable and unbiased monotonic alignment estimation. In addition, we present improved training and inference strategies, including simultaneous fine-tuning from an offline translation model and reduction of monotonic alignment variance. The experimental results demonstrate that the proposed model attains state-of-the-art performance in simultaneous speech-to-text translation on the Spanish and English translation task.

The dynamical properties of skyrmions can be exploited to build devices with new functionalities. Here, we first investigate a skyrmion-based ring-shaped device by means of micromagnetic simulations and Thiele equation. We subsequently show three applications scenarios: (1) a clock with tunable frequency that is biased with an electrical current having a radial spatial distribution, (2) an alternator, where the skyrmion circular motion driven by an engineered anisotropy gradient is converted into an electrical signal, and (3) an energy harvester, where the skyrmion motion driven by a thermal gradient is converted into an electrical signal, thus providing a heat recovery operation. We also show how to precisely tune the frequency and amplitude of the output electrical signals by varying material parameters, geometrical parameters, number and velocity of skyrmions, and we further prove the correct device functionality under realistic conditions given by room temperature and internal material defects. Our results open a new route for the realization of energy efficient nanoscale clocks, generators, and energy harvesters.

What does it take to create the Babel Fish, a tool that can help individuals translate speech between any two languages? While recent breakthroughs in text-based models have pushed machine translation coverage beyond 200 languages, unified speech-to-speech translation models have yet to achieve similar strides. More specifically, conventional speech-to-speech translation systems rely on cascaded systems that perform translation progressively, putting high-performing unified systems out of reach. To address these gaps, we introduce SeamlessM4T, a single model that supports speech-to-speech translation, speech-to-text translation, text-to-speech translation, text-to-text translation, and automatic speech recognition for up to 100 languages. To build this, we used 1 million hours of open speech audio data to learn self-supervised speech representations with w2v-BERT 2.0. Subsequently, we created a multimodal corpus of automatically aligned speech translations. Filtered and combined with human-labeled and pseudo-labeled data, we developed the first multilingual system capable of translating from and into English for both speech and text. On FLEURS, SeamlessM4T sets a new standard for translations into multiple target languages, achieving an improvement of 20% BLEU over the previous SOTA in direct speech-to-text translation. Compared to strong cascaded models, SeamlessM4T improves the quality of into-English translation by 1.3 BLEU points in speech-to-text and by 2.6 ASR-BLEU points in speech-to-speech. Tested for robustness, our system performs better against background noises and speaker variations in speech-to-text tasks compared to the current SOTA model. Critically, we evaluated SeamlessM4T on gender bias and added toxicity to assess translation safety. Finally, all contributions in this work are open-sourced and accessible at https://github.com/facebookresearch/seamless_communication

Modern smartwatches often include photoplethysmographic (PPG) sensors to measure heartbeats or blood pressure through complex algorithms that fuse PPG data with other signals. In this work, we propose a collaborative inference approach that uses both a smartwatch and a connected smartphone to maximize the performance of heart rate (HR) tracking while also maximizing the smartwatch's battery life. In particular, we first analyze the trade-offs between running on-device HR tracking or offloading the work to the mobile. Then, thanks to an additional step to evaluate the difficulty of the upcoming HR prediction, we demonstrate that we can smartly manage the workload between smartwatch and smartphone, maintaining a low mean absolute error (MAE) while reducing energy consumption. We benchmark our approach on a custom smartwatch prototype, including the STM32WB55 MCU and Bluetooth Low-Energy (BLE) communication, and a Raspberry Pi3 as a proxy for the smartphone. With our Collaborative Heart Rate Inference System (CHRIS), we obtain a set of Pareto-optimal configurations demonstrating the same MAE as State-of-Art (SoA) algorithms while consuming less energy. For instance, we can achieve approximately the same MAE of TimePPG-Small (5.54 BPM MAE vs. 5.60 BPM MAE) while reducing the energy by 2.03x, with a configuration that offloads 80\% of the predictions to the phone. Furthermore, accepting a performance degradation to 7.16 BPM of MAE, we can achieve an energy consumption of 179 uJ per prediction, 3.03x less than running TimePPG-Small on the smartwatch, and 1.82x less than streaming all the input data to the phone., Comment: Published at 2023 Design, Automation \& Test in Europe Conference \& Exhibition (DATE)

Magnetic solitons are promising for applications due to their intrinsic properties such as small size, topological stability, ultralow power manipulation and potentially ultrafast operations. To date, research has focused on the manipulation of skyrmions, domain walls, and vortices by applied currents. The discovery of new methods to control magnetic parameters, such as the interfacial Dzyaloshinskii-Moriya interaction (DMI) by strain, geometry design, temperature gradients, and applied voltages promises new avenues for energetically efficient manipulation of magnetic structures. The latter has shown significant progress in 2d material-based technology. In this work, we present a comprehensive study using numerical and analytical methods of the stability and motion of different magnetic textures under the influence of DMI gradients. Our results show that under the influence of linear DMI gradients, N\'eel and Bloch-type skyrmions and radial vortex exhibit motion with finite skyrmion Hall angle, while the circular vortex undergoes expulsion dynamics. This work provides a deeper and crucial understanding of the stability and gradient-driven dynamics of magnetic solitons, and paves the way for the design of alternative low-power sources of magnetization manipulation in the emerging field of 2d materials., Comment: 19 pages, 5 figures

Expanding the language coverage of speech technology has the potential to improve access to information for many more people. However, current speech technology is restricted to about one hundred languages which is a small fraction of the over 7,000 languages spoken around the world. The Massively Multilingual Speech (MMS) project increases the number of supported languages by 10-40x, depending on the task. The main ingredients are a new dataset based on readings of publicly available religious texts and effectively leveraging self-supervised learning. We built pre-trained wav2vec 2.0 models covering 1,406 languages, a single multilingual automatic speech recognition model for 1,107 languages, speech synthesis models for the same number of languages, as well as a language identification model for 4,017 languages. Experiments show that our multilingual speech recognition model more than halves the word error rate of Whisper on 54 languages of the FLEURS benchmark while being trained on a small fraction of the labeled data.

Transducer and Attention based Encoder-Decoder (AED) are two widely used frameworks for speech-to-text tasks. They are designed for different purposes and each has its own benefits and drawbacks for speech-to-text tasks. In order to leverage strengths of both modeling methods, we propose a solution by combining Transducer and Attention based Encoder-Decoder (TAED) for speech-to-text tasks. The new method leverages AED's strength in non-monotonic sequence to sequence learning while retaining Transducer's streaming property. In the proposed framework, Transducer and AED share the same speech encoder. The predictor in Transducer is replaced by the decoder in the AED model, and the outputs of the decoder are conditioned on the speech inputs instead of outputs from an unconditioned language model. The proposed solution ensures that the model is optimized by covering all possible read/write scenarios and creates a matched environment for streaming applications. We evaluate the proposed approach on the \textsc{MuST-C} dataset and the findings demonstrate that TAED performs significantly better than Transducer for offline automatic speech recognition (ASR) and speech-to-text translation (ST) tasks. In the streaming case, TAED outperforms Transducer in the ASR task and one ST direction while comparable results are achieved in another translation direction., Comment: ACL 2023 main conference

We have designed a passive spintronic diode based on a single skyrmion stabilized in a magnetic tunnel junction and studied its dynamics induced by voltage-controlled anisotropy (VCMA) and Dzyaloshinskii-Moriya interaction (VDMI). We have demonstrated that the sensitivity (rectified voltage over input microwave power) with realistic physical parameters and geometry can be larger than 10 kV/W which is one order of magnitude better than diodes employing a uniform ferromagnetic state. Our numerical and analytical results on the VCMA and VDMI-driven resonant excitation of skyrmions beyond the linear regime reveal a frequency dependence on the amplitude and no efficient parametric resonance. Skyrmions with a smaller radius produced higher sensitivities, demonstrating the efficient scalability of skyrmion-based spintronic diodes. These results pave the way for designing passive ultra-sensitive and energy efficient skyrmion-based microwave detectors., Comment: 11 pages, 3 figures

Magnetic skyrmions are fascinating topological particle-like textures promoted by a trade-off among interfacial properties (perpendicular anisotropy and Dzyaloshinskii-Moriya interaction (DMI)) and dipolar interactions. Depending on the dominant interaction, complex spin textures, including pure N\'eel and hybrid skyrmions have been observed in multilayer heterostructures. A quantification of these different spin textures typically requires a depth-reoslved magnetic imaging or scattering techniques. In the present work, we will show qualitatively different collapsing dynamics for pure N\'eel and hybrid skyrmions induced by a perpendicular magnetic field in two representative systems, [Pt/Co/Ir]15 and [Ta/CoFeB/MgO]15 multilayers. Skyrmions in the former stack undergo two morphological transitions, upon reversing the perpendicular field direction. Skyrmions in [Ta/CoFeB/MgO]15 multilayers exhibit a continuous transition, which is mainly linked to a reversible change of the skyrmion size. A full micromagnetic phase diagram is presented to identify these two collapsing mechanisms as a function of material parameters. Since the two distinct collapsing dynamics rely on the detailed layer-dependent spin structures of skyrmions, they could be used as potential fingerprints for identifying the skyrmion type in magnetic multilayers. Our work suggests the employment of pure and hybrid skyrmions for specific applications, due to the strong correlation between the skyrmion dynamics and 3-dimentional spin profiles., Comment: 16 pages, 8 figures

In the "Beyond Moore's Law" era, with increasing edge intelligence, domain-specific computing embracing unconventional approaches will become increasingly prevalent. At the same time, adopting a variety of nanotechnologies will offer benefits in energy cost, computational speed, reduced footprint, cyber resilience, and processing power. The time is ripe for a roadmap for unconventional computing with nanotechnologies to guide future research, and this collection aims to fill that need. The authors provide a comprehensive roadmap for neuromorphic computing using electron spins, memristive devices, two-dimensional nanomaterials, nanomagnets, and various dynamical systems. They also address other paradigms such as Ising machines, Bayesian inference engines, probabilistic computing with p-bits, processing in memory, quantum memories and algorithms, computing with skyrmions and spin waves, and brain-inspired computing for incremental learning and problem-solving in severely resource-constrained environments. These approaches have advantages over traditional Boolean computing based on von Neumann architecture. As the computational requirements for artificial intelligence grow 50 times faster than Moore's Law for electronics, more unconventional approaches to computing and signal processing will appear on the horizon, and this roadmap will help identify future needs and challenges. In a very fertile field, experts in the field aim to present some of the dominant and most promising technologies for unconventional computing that will be around for some time to come. Within a holistic approach, the goal is to provide pathways for solidifying the field and guiding future impactful discoveries., Comment: 80 pages accepted in Nano Futures

With the development of hardware for machine learning, newer models often come at the cost of both increased sizes and computational complexity. In effort to improve the efficiency for these models, we apply and investigate recent quantization techniques on speech representation learning models. The quantization techniques were evaluated on the SUPERB benchmark. On the ASR task, with aggressive quantization to 1 bit, we achieved 86.32% storage reduction (184.42 -> 25.23), 88% estimated runtime reduction (1.00 -> 0.12) with increased word error rate (7.06 -> 15.96). In comparison with DistillHuBERT which also aims for model compression, the 2-bit configuration yielded slightly smaller storage (35.84 vs. 46.98), better word error rate (12.68 vs. 13.37) and more efficient estimated runtime (0.15 vs. 0.73)., Comment: 7 pages

Automatic speech recognition research focuses on training and evaluating on static datasets. Yet, as speech models are increasingly deployed on personal devices, such models encounter user-specific distributional shifts. To simulate this real-world scenario, we introduce LibriContinual, a continual learning benchmark for speaker-specific domain adaptation derived from LibriVox audiobooks, with data corresponding to 118 individual speakers and 6 train splits per speaker of different sizes. Additionally, current speech recognition models and continual learning algorithms are not optimized to be compute-efficient. We adapt a general-purpose training algorithm NetAug for ASR and create a novel Conformer variant called the DisConformer (Disentangled Conformer). This algorithm produces ASR models consisting of a frozen 'core' network for general-purpose use and several tunable 'augment' networks for speaker-specific tuning. Using such models, we propose a novel compute-efficient continual learning algorithm called DisentangledCL. Our experiments show that the DisConformer models significantly outperform baselines on general ASR i.e. LibriSpeech (15.58% rel. WER on test-other). On speaker-specific LibriContinual they significantly outperform trainable-parameter-matched baselines (by 20.65% rel. WER on test) and even match fully finetuned baselines in some settings., Comment: 8 pages, 2 figures. Submitted to ICASSP 2023

We study speech-to-speech translation (S2ST) that translates speech from one language into another language and focuses on building systems to support languages without standard text writing systems. We use English-Taiwanese Hokkien as a case study, and present an end-to-end solution from training data collection, modeling choices to benchmark dataset release. First, we present efforts on creating human annotated data, automatically mining data from large unlabeled speech datasets, and adopting pseudo-labeling to produce weakly supervised data. On the modeling, we take advantage of recent advances in applying self-supervised discrete representations as target for prediction in S2ST and show the effectiveness of leveraging additional text supervision from Mandarin, a language similar to Hokkien, in model training. Finally, we release an S2ST benchmark set to facilitate future research in this field. The demo can be found at https://huggingface.co/spaces/facebook/Hokkien_Translation .