Prateek Verma

Anjalie Field, Prateek Verma, Nay San et al. · stanford

Police body-worn cameras have the potential to improve accountability and transparency in policing. Yet in practice, they result in millions of hours of footage that is never reviewed. We investigate the potential of large pre-trained speech models for facilitating reviews, focusing on ASR and officer speech detection in footage from traffic stops. Our proposed pipeline includes training data alignment and filtering, fine-tuning with resource constraints, and combining officer speech detection with ASR for a fully automated approach. We find that (1) fine-tuning strongly improves ASR performance on officer speech (WER=12-13%), (2) ASR on officer speech is much more accurate than on community member speech (WER=43.55-49.07%), (3) domain-specific tasks like officer speech detection and diarization remain challenging. Our work offers practical applications for reviewing body camera footage and general guidance for adapting pre-trained speech models to noisy multi-speaker domains.

Rishabh Dahale, Vaibhav Talwadker, Preeti Rao et al.

Creating a complex work of art like music necessitates profound creativity. With recent advancements in deep learning and powerful models such as transformers, there has been huge progress in automatic music generation. In an accompaniment generation context, creating a coherent drum pattern with apposite fills and improvisations at proper locations in a song is a challenging task even for an experienced drummer. Drum beats tend to follow a repetitive pattern through stanzas with fills or improvisation at section boundaries. In this work, we tackle the task of drum pattern generation conditioned on the accompanying music played by four melodic instruments: Piano, Guitar, Bass, and Strings. We use the transformer sequence to sequence model to generate a basic drum pattern conditioned on the melodic accompaniment to find that improvisation is largely absent, attributed possibly to its expectedly relatively low representation in the training data. We propose a novelty function to capture the extent of improvisation in a bar relative to its neighbors. We train a model to predict improvisation locations from the melodic accompaniment tracks. Finally, we use a novel BERT-inspired in-filling architecture, to learn the structure of both the drums and melody to in-fill elements of improvised music.

Prateek Verma, Mert Pilanci

Transformer architectures are the backbone of the modern AI revolution. However, they are based on simply stacking the same blocks in dozens of layers and processing information sequentially from one block to another. In this paper, we propose to challenge this and introduce adaptive computations for LLM-like setups, which allow the final layer to attend to all of the intermediate layers as it deems fit through the attention mechanism, thereby introducing computational \textbf{attention shortcuts}. These shortcuts can thus make the architecture depth and context adaptive. We showcase four different datasets, namely acoustic tokens, natural language, and symbolic music, and we achieve superior performance for GPT-like architecture. We give evidence via attention maps that the models learn complex dependencies across layers that are adaptive in context and depth depending on the input tokens.

Prateek Verma, Chris Chafe, Jonathan Berger

The acoustic space in which a sound is created and heard plays an essential role in how that sound is perceived by affording a unique sense of \textit{presence}. Every sound we hear results from successive convolution operations intrinsic to the sound source and external factors such as microphone characteristics and room impulse responses. Typically, researchers use an excitation such as a pistol shot or balloon pop as an impulse signal with which an auralization can be created. The room "impulse" responses convolved with the signal of interest can transform the input sound into the sound played in the acoustic space of interest. Here we propose a novel architecture that can transform a sound of interest into any other acoustic space(room or hall) of interest by using arbitrary audio recorded as a proxy for a balloon pop. The architecture is grounded in simple signal processing ideas to learn residual signals from a learned acoustic signature and the input signal. Our framework allows a neural network to adjust gains of every point in the time-frequency representation, giving sound qualitative and quantitative results.

Prateek Verma, Jonathan Berger

Every sound that we hear is the result of successive convolutional operations (e.g. room acoustics, microphone characteristics, resonant properties of the instrument itself, not to mention characteristics and limitations of the sound reproduction system). In this work we seek to determine the best room in which to perform a particular piece using AI. Additionally, we use room acoustics as a way to enhance the perceptual qualities of a given sound. Historically, rooms (particularly Churches and concert halls) were designed to host and serve specific musical functions. In some cases the architectural acoustical qualities enhanced the music performed there. We try to mimic this, as a first step, by designating room impulse responses that would correlate to producing enhanced sound quality for particular music. A convolutional architecture is first trained to take in an audio sample and mimic the ratings of experts with about 78 % accuracy for various instrument families and notes for perceptual qualities. This gives us a scoring function for any audio sample which can rate the perceptual pleasantness of a note automatically. Now, via a library of about 60,000 synthetic impulse responses mimicking all kinds of room, materials, etc, we use a simple convolution operation, to transform the sound as if it was played in a particular room. The perceptual evaluator is used to rank the musical sounds, and yield the "best room or the concert hall" to play a sound. As a byproduct it can also use room acoustics to turn a poor quality sound into a "good" sound.

Prateek Verma, Chris Chafe

We propose a learnable content adaptive front end for audio signal processing. Before the modern advent of deep learning, we used fixed representation non-learnable front-ends like spectrogram or mel-spectrogram with/without neural architectures. With convolutional architectures supporting various applications such as ASR and acoustic scene understanding, a shift to a learnable front ends occurred in which both the type of basis functions and the weight were learned from scratch and optimized for the particular task of interest. With the shift to transformer-based architectures with no convolutional blocks present, a linear layer projects small waveform patches onto a small latent dimension before feeding them to a transformer architecture. In this work, we propose a way of computing a content-adaptive learnable time-frequency representation. We pass each audio signal through a bank of convolutional filters, each giving a fixed-dimensional vector. It is akin to learning a bank of finite impulse-response filterbanks and passing the input signal through the optimum filter bank depending on the content of the input signal. A content-adaptive learnable time-frequency representation may be more broadly applicable, beyond the experiments in this paper.

Prateek Verma, Shan Zhong, Xiaoyu Liu et al.

Query autocomplete (QAC) also known as typeahead, suggests list of complete queries as user types prefix in the search box. It is one of the key features of modern search engines specially in e-commerce. One of the goals of typeahead is to suggest relevant queries to users which are seasonally important. In this paper we propose a neural network based natural language processing (NLP) algorithm to incorporate seasonality as a signal and present end to end evaluation of the QAC ranking model. Incorporating seasonality into autocomplete ranking model can improve autocomplete relevance and business metric.

Prateek Verma

This work builds together two popular blocks of neural architecture, namely convolutional layers and Transformers, for large language models (LLMs). Non-causal conformers are used ubiquitously in automatic speech recognition. This work aims to adapt these architectures in a causal setup for training LLMs. Transformers decoders effectively capture long-range dependencies over several modalities and form a core backbone of modern advancements in machine learning. Convolutional architectures have been popular in extracting features in domains such as raw 1-D signals, speech, and images, to name a few. In this paper, by combining local and global dependencies over latent representations using causal convolutional filters and Transformer, we achieve significant gains in performance. This work showcases a robust speech architecture that can be integrated and adapted in a causal setup beyond speech applications for large-scale language modeling.

Prateek Verma

With the advent of modern AI architectures, a shift has happened towards end-to-end architectures. This pivot has led to neural architectures being trained without domain-specific biases/knowledge, optimized according to the task. We in this paper, learn audio embeddings via diverse feature representations, in this case, domain-specific. For the case of audio classification over hundreds of categories of sound, we learn robust separate embeddings for diverse audio properties such as pitch, timbre, and neural representation, along with also learning it via an end-to-end architecture. We observe handcrafted embeddings, e.g., pitch and timbre-based, although on their own, are not able to beat a fully end-to-end representation, yet adding these together with end-to-end embedding helps us, significantly improve performance. This work would pave the way to bring some domain expertise with end-to-end models to learn robust, diverse representations, surpassing the performance of just training end-to-end models.

Prateek Verma

This report will explore and answer fundamental questions about taking Fourier Transforms and tying it with recent advances in AI and neural architecture. One interpretation of the Fourier Transform is decomposing a signal into its constituent components by projecting them onto complex exponentials. Variants exist, such as discrete cosine transform that does not operate on the complex domain and projects an input signal to only cosine functions oscillating at different frequencies. However, this is a fundamental limitation, and it needs to be more suboptimal. The first one is that all kernels are sinusoidal: What if we could have some kernels adapted or learned according to the problem? What if we can use neural architectures for this? We show how one can learn these kernels from scratch for audio signal processing applications. We find that the neural architecture not only learns sinusoidal kernel shapes but discovers all kinds of incredible signal-processing properties. E.g., windowing functions, onset detectors, high pass filters, low pass filters, modulations, etc. Further, upon analysis of the filters, we find that the neural architecture has a comb filter-like structure on top of the learned kernels. Comb filters that allow harmonic frequencies to pass through are one of the core building blocks/types of filters similar to high-pass, low-pass, and band-pass filters of various traditional signal processing algorithms. Further, we can also use the convolution operation with a signal to be learned from scratch, and we will explore papers in the literature that uses this with that robust Transformer architectures. Further, we would also explore making the learned kernel's content adaptive, i.e., learning different kernels for different inputs.

Prateek Verma

Modeling long-term dependencies for audio signals is a particularly challenging problem, as even small-time scales yield on the order of a hundred thousand samples. With the recent advent of Transformers, neural architectures became good at modeling dependencies over longer time scales, but they suffered from quadratic constraints to scale them. We propose a generative auto-regressive architecture that can model audio waveforms over quite a large context, greater than 500,000 samples. Our work is adapted to learn time dependencies by learning a latent representation by a CNN front-end, and then learning dependencies over these representations using Transformer encoders, fully trained end-to-end: thereby allowing to learn representations as it deems fit for the next sample. Unlike previous works that compared different time scales to show improvement, we use a standard dataset, with the same number of parameters/context to show improvements. We achieve a state-of-the-art performance as compared to other approaches such as Wavenet, SaSHMI, and Sample-RNN on a standard dataset for modeling long-term structure. This work gives very exciting direction for the field, given improvements in context modeling that can be scaled with more data, as well as potentially better results by using billions/trillions of parameters.

Minh-Hao Van, Prateek Verma, Chen Zhao et al.

Foundation models (FMs) are catalyzing a transformative shift in materials science (MatSci) by enabling scalable, general-purpose, and multimodal AI systems for scientific discovery. Unlike traditional machine learning models, which are typically narrow in scope and require task-specific engineering, FMs offer cross-domain generalization and exhibit emergent capabilities. Their versatility is especially well-suited to materials science, where research challenges span diverse data types and scales. This survey provides a comprehensive overview of foundation models, agentic systems, datasets, and computational tools supporting this growing field. We introduce a task-driven taxonomy encompassing six broad application areas: data extraction, interpretation and Q\&A; atomistic simulation; property prediction; materials structure, design and discovery; process planning, discovery, and optimization; and multiscale modeling. We discuss recent advances in both unimodal and multimodal FMs, as well as emerging large language model (LLM) agents. Furthermore, we review standardized datasets, open-source tools, and autonomous experimental platforms that collectively fuel the development and integration of FMs into research workflows. We assess the early successes of foundation models and identify persistent limitations, including challenges in generalizability, interpretability, data imbalance, safety concerns, and limited multimodal fusion. Finally, we articulate future research directions centered on scalable pretraining, continual learning, data governance, and trustworthiness.

An Vuong, Minh-Hao Van, Prateek Verma et al.

Vision-Language Models (VLMs) have shown strong performance in tasks like visual question answering and multimodal text generation, but their effectiveness in scientific domains such as materials science remains limited. While some machine learning methods have addressed specific challenges in this field, there is still a lack of foundation models designed for broad tasks like polymer property prediction using multimodal data. In this work, we present a multimodal polymer dataset to fine-tune VLMs through instruction-tuning pairs and assess the impact of multimodality on prediction performance. Our fine-tuned models, using LoRA, outperform unimodal and baseline approaches, demonstrating the benefits of multimodal learning. Additionally, this approach reduces the need to train separate models for different properties, lowering deployment and maintenance costs.

Prateek Verma, Mert Pilanci

We propose a framework that enables neural models to "think while listening" to everyday sounds, thereby enhancing audio classification performance. Motivated by recent advances in the reasoning capabilities of large language models, we address two central questions: (i) how can thinking be incorporated into existing audio classification pipelines to enable reasoning in the category space and improve performance, and (ii) can a new architecture be designed from the ground up to support both thinking and test-time scaling? We demonstrate that in both settings, our models exhibit improved classification accuracy. Leveraging test-time scaling, we observe consistent gains as the number of sampled traces increases. Furthermore, we evaluate two open-source reasoning models, GPT-OSS-20B and Qwen3-14B, showing that while such models are capable of zero-shot reasoning, a lightweight approach--retraining only the embedding matrix of a frozen, smaller model like GPT-2--can surpass the performance of billion-parameter text-based reasoning models.

Prateek Verma

Large Language Models (LLMs) have ushered in a new wave of artificial intelligence advancements impacting every scientific field and discipline. We live in a world where most of the data around us, e.g., text, audio, and music, has a multi-scale structure. This paper infuses LLMs with a traditional signal processing idea, namely wavelets, during pre-training to take advantage of the structure. Without adding \textbf{any extra parameters} to a GPT-style LLM architecture in an academic setup, we achieve the same pre-training performance almost twice as fast in text, audio, and images. This is done by imposing a structure on intermediate embeddings. When trained for the same number of training steps, we achieve significant gains in performance, which is comparable to pre-training a larger neural architecture. Further, we show this extends to the Long Range Arena benchmark and several input representations such as characters, BPE tokens, bytes, waveform, math expression, and image pixels. Our architecture allows every next token prediction access to intermediate embeddings at different temporal resolutions in every decoder block. We hope this will pave the way for incorporating multi-rate signal processing into pre-training.

Minh-Hao Van, Prateek Verma, Xintao Wu

Recently, large language models (LLMs) have taken the spotlight in natural language processing. Further, integrating LLMs with vision enables the users to explore emergent abilities with multimodal data. Visual language models (VLMs), such as LLaVA, Flamingo, or CLIP, have demonstrated impressive performance on various visio-linguistic tasks. Consequently, there are enormous applications of large models that could be potentially used in the biomedical imaging field. Along that direction, there is a lack of related work to show the ability of large models to diagnose the diseases. In this work, we study the zero-shot and few-shot robustness of VLMs on the medical imaging analysis tasks. Our comprehensive experiments demonstrate the effectiveness of VLMs in analyzing biomedical images such as brain MRIs, microscopic images of blood cells, and chest X-rays.

Prateek Verma, Minh-Hao Van, Xintao Wu

Vision language models (VLMs) have recently emerged and gained the spotlight for their ability to comprehend the dual modality of image and textual data. VLMs such as LLaVA, ChatGPT-4, and Gemini have recently shown impressive performance on tasks such as natural image captioning, visual question answering (VQA), and spatial reasoning. Additionally, a universal segmentation model by Meta AI, Segment Anything Model (SAM) shows unprecedented performance at isolating objects from unforeseen images. Since medical experts, biologists, and materials scientists routinely examine microscopy or medical images in conjunction with textual information in the form of captions, literature, or reports, and draw conclusions of great importance and merit, it is indubitably essential to test the performance of VLMs and foundation models such as SAM, on these images. In this study, we charge ChatGPT, LLaVA, Gemini, and SAM with classification, segmentation, counting, and VQA tasks on a variety of microscopy images. We observe that ChatGPT and Gemini are impressively able to comprehend the visual features in microscopy images, while SAM is quite capable at isolating artefacts in a general sense. However, the performance is not close to that of a domain expert - the models are readily encumbered by the introduction of impurities, defects, artefact overlaps and diversity present in the images.

Prateek Verma, Mert Pilanci

This paper presents a fascinating find: By training an auto-regressive LLM model on text tokens, the text model inherently develops internally an ability to understand images and audio, thereby developing the ability to see and hear just by reading. Popular audio and visual LLM models fine-tune text LLM models to give text output conditioned on images and audio embeddings. On the other hand, our architecture takes in patches of images, audio waveforms or tokens as input. It gives us the embeddings or category labels typical of a classification pipeline. We show the generality of text weights in aiding audio classification for datasets FSD-50K and GTZAN. Further, we show this working for image classification on CIFAR-10 and Fashion-MNIST, as well on image patches. This pushes the notion of text-LLMs learning powerful internal circuits that can be utilized by activating necessary connections for various applications rather than training models from scratch every single time.

Adithya Rajan, Xiaoyu Liu, Prateek Verma et al.

We introduce a data-centric approach for mitigating presentation bias in real-time neural query autocomplete systems through the use of synthetic prefixes. These prefixes are generated from complete user queries collected during regular search sessions where autocomplete was not active. This allows us to enrich the training data for learning to rank models with more diverse and less biased examples. This method addresses the inherent bias in engagement signals collected from live query autocomplete interactions, where model suggestions influence user behavior. Our neural ranker is optimized for real-time deployment under strict latency constraints and incorporates a rich set of features, including query popularity, seasonality, fuzzy match scores, and contextual signals such as department affinity, device type, and vertical alignment with previous user queries. To support efficient training, we introduce a task-specific simplification of the listwise loss, reducing computational complexity from $O(n^2)$ to $O(n)$ by leveraging the query autocomplete structure of having only one ground-truth selection per prefix. Deployed in a large-scale e-commerce setting, our system demonstrates statistically significant improvements in user engagement, as measured by mean reciprocal rank and related metrics. Our findings show that synthetic prefixes not only improve generalization but also provide a scalable path toward bias mitigation in other low-latency ranking tasks, including related searches and query recommendations.

Prateek Verma

We propose WHISPER-GPT: A generative large language model (LLM) for speech and music that allows us to work with continuous audio representations and discrete tokens simultaneously as part of a single architecture. There has been a huge surge in generative audio, speech, and music models that utilize discrete audio tokens derived from neural compression algorithms, e.g. ENCODEC. However, one of the major drawbacks of this approach is handling the context length. It blows up for high-fidelity generative architecture if one has to account for all the audio contents at various frequencies for the next token prediction. By combining continuous audio representation like the spectrogram and discrete acoustic tokens, we retain the best of both worlds: Have all the information needed from the audio at a specific time instance in a single token, yet allow LLM to predict the future token to allow for sampling and other benefits discrete space provides. We show how our architecture improves the perplexity and negative log-likelihood scores for the next token prediction compared to a token-based LLM for speech and music.

Prateek Verma, Mert Pilanci

This paper introduces the idea of applying signal processing inside a Large Language Model (LLM). With the recent explosion of generative AI, our work can help bridge two fields together, namely the field of signal processing and large language models. We draw parallels between classical Fourier-Transforms and Fourier Transform-like learnable time-frequency representations for every intermediate activation signal of an LLM. Once we decompose every activation signal across tokens into a time-frequency representation, we learn how to filter and reconstruct them, with all components learned from scratch, to predict the next token given the previous context. We show that for GPT-like architectures, our work achieves faster convergence and significantly increases performance by adding a minuscule number of extra parameters when trained for the same epochs. We hope this work paves the way for algorithms exploring signal processing inside the signals found in neural architectures like LLMs and beyond.

Prateek Verma

This paper presents a way of doing large scale audio understanding without traditional state of the art neural architectures. Ever since the introduction of deep learning for understanding audio signals in the past decade, convolutional architectures have been able to achieve state of the art results surpassing traditional hand-crafted features. In the recent past, there has been a similar shift away from traditional convolutional and recurrent neural networks towards purely end-to-end Transformer architectures. We, in this work, explore an approach, based on Bag-of-Words model. Our approach does not have any convolutions, recurrence, attention, transformers or other approaches such as BERT. We utilize micro and macro level clustered vanilla embeddings, and use a MLP head for classification. We only use feed-forward encoder-decoder models to get the bottlenecks of spectral envelops, spectral patches and slices as well as multi-resolution spectra. A classification head (a feed-forward layer), similar to the approach in SimCLR is trained on a learned representation. Using simple codes learned on latent representations, we show how we surpass traditional convolutional neural network architectures, and come strikingly close to outperforming powerful Transformer architectures. This work hopefully would pave way for exciting advancements in the field of representation learning without massive, end-to-end neural architectures.

Prateek Verma, Chris Chafe

This paper proposes a novel way of doing audio synthesis at the waveform level using Transformer architectures. We propose a deep neural network for generating waveforms, similar to wavenet. This is fully probabilistic, auto-regressive, and causal, i.e. each sample generated depends only on the previously observed samples. Our approach outperforms a widely used wavenet architecture by up to 9% on a similar dataset for predicting the next step. Using the attention mechanism, we enable the architecture to learn which audio samples are important for the prediction of the future sample. We show how causal transformer generative models can be used for raw waveform synthesis. We also show that this performance can be improved by another 2% by conditioning samples over a wider context. The flexibility of the current model to synthesize audio from latent representations suggests a large number of potential applications. The novel approach of using generative transformer architectures for raw audio synthesis is, however, still far away from generating any meaningful music, without using latent codes/meta-data to aid the generation process.

Prateek Verma, Jonathan Berger

Over the past two decades, CNN architectures have produced compelling models of sound perception and cognition, learning hierarchical organizations of features. Analogous to successes in computer vision, audio feature classification can be optimized for a particular task of interest, over a wide variety of datasets and labels. In fact similar architectures designed for image understanding have proven effective for acoustic scene analysis. Here we propose applying Transformer based architectures without convolutional layers to raw audio signals. On a standard dataset of Free Sound 50K,comprising of 200 categories, our model outperforms convolutional models to produce state of the art results. This is significant as unlike in natural language processing and computer vision, we do not perform unsupervised pre-training for outperforming convolutional architectures. On the same training set, with respect mean aver-age precision benchmarks, we show a significant improvement. We further improve the performance of Transformer architectures by using techniques such as pooling inspired from convolutional net-work designed in the past few years. In addition, we also show how multi-rate signal processing ideas inspired from wavelets, can be applied to the Transformer embeddings to improve the results. We also show how our models learns a non-linear non constant band-width filter-bank, which shows an adaptable time frequency front end representation for the task of audio understanding, different from other tasks e.g. pitch estimation.

Bibit Bianchini, Prateek Verma, Kenneth Salisbury

Physical human-robot interactions (pHRI) are less efficient and communicative than human-human interactions, and a key reason is a lack of informative sense of touch in robotic systems. Interpreting human touch gestures is a nuanced, challenging task with extreme gaps between human and robot capability. Among prior works that demonstrate human touch recognition capability, differences in sensors, gesture classes, feature sets, and classification algorithms yield a conglomerate of non-transferable results and a glaring lack of a standard. To address this gap, this work presents 1) four proposed touch gesture classes that cover an important subset of the gesture characteristics identified in the literature, 2) the collection of an extensive force dataset on a common pHRI robotic arm with only its internal wrist force-torque sensor, and 3) an exhaustive performance comparison of combinations of feature sets and classification algorithms on this dataset. We demonstrate high classification accuracies among our proposed gesture definitions on a test set, emphasizing that neural net-work classifiers on the raw data outperform other combinations of feature sets and algorithms. The accompanying video is here: https://youtu.be/gJPVImNKU68

Prateek Verma, Julius Smith

In this paper, we present a framework for contrastive learning for audio representations, in a self supervised frame work without access to any ground truth labels. The core idea in self supervised contrastive learning is to map an audio signal and its various augmented versions (representative of salient aspects of audio like pitch, timbre etc.) to a space where they are close together, and are separated from other different signals. In addition we also explore generative models based on state of the art transformer based architectures for learning latent spaces for audio signals, without access to any labels. Here, we map audio signals on a smaller scale to discrete dictionary elements and train transformers to predict the next dictionary element. We only use data as a method of supervision, bypassing the need of labels needed to act as a supervision for training the deep neural networks. We then use a linear classifier head in order to evaluate the performance of our models, for both self supervised contrastive and generative transformer based representations that are learned. Our system achieves considerable performance, compared to a fully supervised method, with access to ground truth labels to train the neural network model. These representations, with avail-ability of large scale audio data show promise in various tasks for audio understanding tasks

Prateek Verma, Constantin Basica, Pamela Davis Kivelson

We propose a system that learns from artistic pairings of music and corresponding album cover art. The goal is to 'translate' paintings into music and, in further stages of development, the converse. We aim to deploy this system as an artistic tool for real time 'translations' between musicians and painters. The system's outputs serve as elements to be employed in a joint live performance of music and painting, or as generative material to be used by the artists as inspiration for their improvisation.

Camille Noufi, Prateek Verma

In music and speech, meaning is derived at multiple levels of context. Affect, for example, can be inferred both by a short sound token and by sonic patterns over a longer temporal window such as an entire recording. In this letter, we focus on inferring meaning from this dichotomy of contexts. We show how contextual representations of short sung vocal lines can be implicitly learned from fundamental frequency ($F_0$) and thus be used as a meaningful feature space for downstream Music Information Retrieval (MIR) tasks. We propose three self-supervised deep learning paradigms which leverage pseudotask learning of these two levels of context to produce latent representation spaces. We evaluate the usefulness of these representations by embedding unseen pitch contours into each space and conducting downstream classification tasks. Our results show that contextual representation can enhance downstream classification by as much as 15\% as compared to using traditional statistical contour features.

Prateek Verma, Alessandro Ilic Mezza, Chris Chafe et al.

Networked Music Performance (NMP) is envisioned as a potential game changer among Internet applications: it aims at revolutionizing the traditional concept of musical interaction by enabling remote musicians to interact and perform together through a telecommunication network. Ensuring realistic conditions for music performance, however, constitutes a significant engineering challenge due to extremely strict requirements in terms of audio quality and, most importantly, network delay. To minimize the end-to-end delay experienced by the musicians, typical implementations of NMP applications use un-compressed, bidirectional audio streams and leverage UDP as transport protocol. Being connection less and unreliable,audio packets transmitted via UDP which become lost in transit are not re-transmitted and thus cause glitches in the receiver audio playout. This article describes a technique for predicting lost packet content in real-time using a deep learning approach. The ability of concealing errors in real time can help mitigate audio impairments caused by packet losses, thus improving the quality of audio playout in real-world scenarios.

Prateek Verma, Kenneth Salisbury

Audio perception is a key to solving a variety of problems ranging from acoustic scene analysis, music meta-data extraction, recommendation, synthesis and analysis. It can potentially also augment computers in doing tasks that humans do effortlessly in day-to-day activities. This paper builds upon key ideas to build perception of touch sounds without access to any ground-truth data. We show how we can leverage ideas from classical signal processing to get large amounts of data of any sound of interest with a high precision. These sounds are then used, along with the images to map the sounds to a clustered space of the latent representation of these images. This approach, not only allows us to learn semantic representation of the possible sounds of interest, but also allows association of different modalities to the learned distinctions. The model trained to map sounds to this clustered representation, gives reasonable performance as opposed to expensive methods collecting a lot of human annotated data. Such approaches can be used to build a state of art perceptual model for any sound of interest described using a few signal processing features. Daisy chaining high precision sound event detectors using signal processing combined with neural architectures and high dimensional clustering of unlabelled data is a vastly powerful idea, and can be explored in a variety of ways in future.

Prateek Verma, Jonathan Berger

We present a framework to model the perceived quality of audio signals by combining convolutional architectures, with ideas from classical signal processing, and describe an approach to enhancing perceived acoustical quality. We demonstrate the approach by transforming the sound of an inexpensive musical with degraded sound quality to that of a high-quality musical instrument without the need for parallel data which is often hard to collect. We adapt the classical approach of a simple adaptive EQ filtering to the objective criterion learned by a neural architecture and optimize it to get the signal of our interest. Since we learn adaptive masks depending on the signal of interest as opposed to a fixed transformation for all the inputs, we show that shallow neural architectures can achieve the desired result. A simple constraint on the objective and the initialization helps us in avoiding adversarial examples, which otherwise would have produced noisy, unintelligible audio. We believe that the current framework proposed has enormous applications, in a variety of problems where one can learn a loss function depending on the problem, using a neural architecture and optimize it after it has been learned.

Prateek Verma, Aliasgar Kutiyanawala, Ke Shen

Products in an ecommerce catalog contain information-rich fields like description and bullets that can be useful to extract entities (attributes) using NER based systems. However, these fields are often verbose and contain lot of information that is not relevant from a search perspective. Treating each sentence within these fields equally can lead to poor full text match and introduce problems in extracting attributes to develop ontologies, semantic search etc. To address this issue, we describe two methods based on extractive summarization with reinforcement learning by leveraging information in product titles and search click through logs to rank sentences from bullets, description, etc. Finally, we compare the accuracy of these two models.

Amruta Vidwans, Prateek Verma, Preeti Rao

We present a melody based classification of musical styles by exploiting the pitch and energy based characteristics derived from the audio signal. Three prominent musical styles were chosen which have improvisation as integral part with similar melodic principles, theme, and structure of concerts namely, Hindustani, Carnatic and Turkish music. Listeners of one or more of these genres can discriminate between these based on the melodic contour alone. Listening tests were carried out using melodic attributes alone, on similar melodic pieces with respect to raga/makam, and removing any instrumentation cue to validate our hypothesis that style distinction is evident in the melody. Our method is based on finding a set of highly discriminatory features, derived from musicology, to capture distinct characteristics of the melodic contour. Behavior in terms of transitions of the pitch contour, the presence of micro-tonal notes and the nature of variations in the vocal energy are exploited. The automatically classified style labels are found to correlate well with subjective listening judgments. This was verified by using statistical tests to compare the labels from subjective and objective judgments. The melody based features, when combined with timbre based features, were seen to improve the classification performance.

Michelle Guo, Albert Haque, Prateek Verma

In this paper, we address the task of spoken language understanding. We present a method for translating spoken sentences from one language into spoken sentences in another language. Given spectrogram-spectrogram pairs, our model can be trained completely from scratch to translate unseen sentences. Our method consists of a pyramidal-bidirectional recurrent network combined with a convolutional network to output sentence-level spectrograms in the target language. Empirically, our model achieves competitive performance with state-of-the-art methods on multiple languages and can generalize to unseen speakers.

Prateek Verma, Chris Chafe, Jonathan Berger

We propose the Neuralogram -- a deep neural network based representation for understanding audio signals which, as the name suggests, transforms an audio signal to a dense, compact representation based upon embeddings learned via a neural architecture. Through a series of probing signals, we show how our representation can encapsulate pitch, timbre and rhythm-based information, and other attributes. This representation suggests a method for revealing meaningful relationships in arbitrarily long audio signals that are not readily represented by existing algorithms. This has the potential for numerous applications in audio understanding, music recommendation, meta-data extraction to name a few.

Albert Haque, Michelle Guo, Prateek Verma et al.

We propose spoken sentence embeddings which capture both acoustic and linguistic content. While existing works operate at the character, phoneme, or word level, our method learns long-term dependencies by modeling speech at the sentence level. Formulated as an audio-linguistic multitask learning problem, our encoder-decoder model simultaneously reconstructs acoustic and natural language features from audio. Our results show that spoken sentence embeddings outperform phoneme and word-level baselines on speech recognition and emotion recognition tasks. Ablation studies show that our embeddings can better model high-level acoustic concepts while retaining linguistic content. Overall, our work illustrates the viability of generic, multi-modal sentence embeddings for spoken language understanding.

Aliasgar Kutiyanawala, Prateek Verma, Zheng et al.

Query Understanding is a semantic search method that can classify tokens in a customer's search query to entities such as Product, Brand, etc. This method can overcome the limitations of bag-of-words methods but requires an ontology. We show that current ontologies are not optimized for search and propose a simplified ontology framework designed specifically for e-commerce search and retrieval. We also present three methods for automatically extracting product classes for the proposed ontology and compare their performance relative to each other.

Albert Haque, Michelle Guo, Prateek Verma

We present an end-to-end method for transforming audio from one style to another. For the case of speech, by conditioning on speaker identities, we can train a single model to transform words spoken by multiple people into multiple target voices. For the case of music, we can specify musical instruments and achieve the same result. Architecturally, our method is a fully-differentiable sequence-to-sequence model based on convolutional and hierarchical recurrent neural networks. It is designed to capture long-term acoustic dependencies, requires minimal post-processing, and produces realistic audio transforms. Ablation studies confirm that our model can separate speaker and instrument properties from acoustic content at different receptive fields. Empirically, our method achieves competitive performance on community-standard datasets.

Prateek Verma, Julius O. Smith

There has been fascinating work on creating artistic transformations of images by Gatys. This was revolutionary in how we can in some sense alter the 'style' of an image while generally preserving its 'content'. In our work, we present a method for creating new sounds using a similar approach, treating it as a style-transfer problem, starting from a random-noise input signal and iteratively using back-propagation to optimize the sound to conform to filter-outputs from a pre-trained neural architecture of interest. For demonstration, we investigate two different tasks, resulting in bandwidth expansion/compression, and timbral transfer from singing voice to musical instruments. A feature of our method is that a single architecture can generate these different audio-style-transfer types using the same set of parameters which otherwise require different complex hand-tuned diverse signal processing pipelines.