Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

Digital processing of speech signals

An intelligent automated assistant system engages with the user in an integrated, conversational manner using natural language dialog, and invokes external services when appropriate to obtain information or perform various actions. The system can be implemented using any of a number of different platforms, such as the web, email, smartphone, and the like, or any combination thereof. In one embodiment, the system is based on sets of interrelated domains and tasks, and employs additional functionally powered by external services with which the system can interact.

Intelligent Automated Assistant

When building a unified vision system or gradually adding new capabilities to a system, the usual assumption is that training data for all tasks is always available. However, as the number of tasks grows, storing and retraining on such data becomes infeasible. A new problem arises where we add new capabilities to a Convolutional Neural Network (CNN), but the training data for its existing capabilities are unavailable. We propose our Learning without Forgetting method, which uses only new task data to train the network while preserving the original capabilities. Our method performs favorably compared to commonly used feature extraction and fine-tuning adaption techniques and performs similarly to multitask learning that uses original task data we assume unavailable. A more surprising observation is that Learning without Forgetting may be able to replace fine-tuning with similar old and new task datasets for improved new task performance.

Learning without Forgetting

A driving function relying on map data to operate is prone to failures that are due to deviations between the real world and the map data. Hence, we transfer the concept of dependability known from systems engineering to high-definition (HD) maps to allow for a system ensuring three major aspects of dependability: reliability, availability, and the safe use of map data within the vehicle. In this paper, we therefore define a coherent terminology in the field, particularly introducing necessary terms for describing and measuring map deviations. To substantiate our terminology, we present the results of a measurement campaign and analyze map deviations in an urban domain HD map after a period of 2 years. The results show that the analyzed map contains relatively few errors (0.07 / km) and no obvious persistent changes over the years but is subject to a comparatively high number of temporary changes (0.74 / km) rendering almost 9% of the map outdated.

Terminology and Analysis of Map Deviations in Urban Domains: Towards Dependability for HD Maps in Automated Vehicles.

A microphone array beamformer combined with a post-filter estimation based on spatial smoothing can deliver good noise attenuation preserving the speech component, however, with disturbing musical tones. On the other hand, the temporal smoothing of the decision-directed (DD) a priori signal-to-noise ratio (SNR) estimation for single-channel noise reduction can suppress musical tones well, however, with speech distortion particularly in speech onset. Based on these facts, we derive a new data-driven multichannel a priori SNR estimation based on both spatial and temporal smoothing for the use in a beamformer post-filter. The new a priori SNR estimation is able to find an optimum compromise between noise attenuation, quality of the speech component, and musical tones suppression.

A data-driven post-filter design based on spatially and temporally smoothed a priori SNR

Automatic speech recognition (ASR) for wideband (WB) telephone speech services must cope with a lack of matching speech databases for acoustic model training. This paper investigates the impact of mixing insufficient WB and additional narrowband (NB) speech training data. It turns out that decimation and interpolation techniques, reducing the bandwidth mismatch between the NB speech material in training and the WB speech data to be recognized, do not succeed in outperforming the pure NB ASR baseline. However, true WBASR training supported by artificial bandwidth extension (ABE) reveals a performance gain. A new ABE approach that makes use of robust dynamic features and a Viterbi path decoder exploiting phonetic a priori knowledge proves to be superior. It yields a reduction of 1.9 % word error rate relative to the NB ASR baseline and 9.3 % relative to a WB ASR experiment trained on only a limited amount of WB speech data.

Automatic recognition of wideband telephone speech with limited amount of matched training data

Microphone arrays and beamforming for car applications have gained much attention in the past. While several proposals aim at quite idealistic conditions, our approach intentionally employs low-cost microphones. Moreover, we assume the microphone array to be integrated into the head-unit of the car. Although from an acoustical perspective this is not an ideal location, it is yet very attractive since it omits further wiring and offers a compact and optimized solution. To achieve the required level of directivity and noise reduction in car noise, the a priori noise field of diffuse noise is exploited. A new postfilter estimation using an adaptive smoothing factor along with an estimation quality indicator is proposed which is well suited for low-cost microphones. Meanwhile, an intrusive instrumental evaluation methodology will be introduced. We will show that a significant level of noise attenuation can be achieved, while the quality of the speech component will be improved compared to the state-of-the-art.

A post-filter for wideband speech beamforming in automotive application

In anticipation of upcoming mobile telephony services with higher speech quality, a wideband (50 Hz to 7 kHz) mobile telephony derivative of TIMIT has been recorded called WTIMIT It opens up various scientific investigations; eg, on speech quality and intelligibility, as well as on wideband upgrades of network-side interactive voice response (IVR) systems with retrained or bandwidth-extended acoustic models for automatic speech recognition (ASR) Wideband telephony could enable network-side speech recognition applications such as remote dictation or spelling without the need of distributed speech recognition techniques The WTIMIT corpus was transmitted via two prepared Nokia 6220 mobile phones over T-Mobile's 3G wideband mobile network in The Hague, The Netherlands, employing the Adaptive Multirate Wideband (AMR-WB) speech codec The paper presents observations of transmission effects and phoneme recognition experiments It turns out that in the case of wideband telephony, server-side ASR should not be carried out by simply decimating received signals to 8 kHz and applying existent narrowband acoustic models Nor do we recommend just simulating the AMR-WB codec for training of wideband acoustic models Instead, real-world wideband telephony channel data (such as WTIMIT) provides the best training material for wideband IVR systems

/pdf/wtimit-the-timit-speech-corpus-transmitted-over-the-3g-amr-7h4o6njaa2.pdf

Tim Fingscheidt

Papers

Terminology and Analysis of Map Deviations in Urban Domains: Towards Dependability for HD Maps in Automated Vehicles.

A data-driven post-filter design based on spatially and temporally smoothed a priori SNR

Automatic recognition of wideband telephone speech with limited amount of matched training data

A post-filter for wideband speech beamforming in automotive application

WTIMIT: The TIMIT Speech Corpus Transmitted Over The 3G AMR Wideband Mobile Network