Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userFast Development of ASR in African Languages using Self Supervised Speech Representation Learning
274
0
0.0
(
0
)
Added by
Laurent Besacier
Publication date
2021
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
This paper describes the results of an informal collaboration launched during the African Master of Machine Intelligence (AMMI) in June 2020. After a series of lectures and labs on speech data collection using mobile applications and on self-supervised representation learning from speech, a small group of students and the lecturer continued working on automatic speech recognition (ASR) project for three languages: Wolof, Ga, and Somali. This paper describes how data was collected and ASR systems developed with a small amount (1h) of transcribed speech as training data. In these low resource conditions, pre-training a model on large amounts of raw speech was fundamental for the efficiency of ASR systems developed.
rate research
Read More
This paper presents a self-supervised learning framework, named MGF, for general-purpose speech representation learning. In the design of MGF, speech hierarchy is taken into consideration. Specifically, we propose to use generative learning approaches to capture fine-grained information at small time scales and use discriminative learning approaches to distill coarse-grained or semantic information at large time scales. For phoneme-scale learning, we borrow idea from the masked language model but tailor it for the continuous speech signal by replacing classification loss with a contrastive loss. We corroborate our design by evaluating MGF representation on various downstream tasks, including phoneme classification, speaker classification, speech recognition, and emotion classification. Experiments verify that training at different time scales needs different training targets and loss functions, which in general complement each other and lead to a better performance.
In order to evaluate the performance of the attention based neural ASR under noisy conditions, the current trend is to present hours of various noisy speech data to the model and measure the overall word/phoneme error rate (W/PER). In general, it is unclear how these models perform when exposed to a cocktail party setup in which two or more speakers are active. In this paper, we present the mixtures of speech signals to a popular attention-based neural ASR, known as Listen, Attend, and Spell (LAS), at different target-to-interference ratio (TIR) and measure the phoneme error rate. In particular, we investigate in details when two phonemes are mixed what will be the predicted phoneme; in this fashion we build a model in which the most probable predictions for a phoneme are given. We found a 65% relative increase in PER when LAS was presented with mixed speech signals at TIR = 0 dB and the performance approaches the unmixed scenario at TIR = 30 dB. Our results show the model, when presented with mixed phonemes signals, tend to predict those that have higher accuracies during evaluation of original phoneme signals.
End-to-end neural network models achieve improved performance on various automatic speech recognition (ASR) tasks. However, these models perform poorly on edge hardware due to large memory and computation requirements. While quantizing model weights and/or activations to low-precision can be a promising solution, previous research on quantizing ASR models is limited. Most quantization approaches use floating-point arithmetic during inference; and thus they cannot fully exploit integer processing units, which use less power than their floating-point counterparts. Moreover, they require training/validation data during quantization for finetuning or calibration; however, this data may not be available due to security/privacy concerns. To address these limitations, we propose Q-ASR, an integer-only, zero-shot quantization scheme for ASR models. In particular, we generate synthetic data whose runtime statistics resemble the real data, and we use it to calibrate models during quantization. We then apply Q-ASR to quantize QuartzNet-15x5 and JasperDR-10x5 without any training data, and we show negligible WER change as compared to the full-precision baseline models. For INT8-only quantization, we observe a very modest WER degradation of up to 0.29%, while we achieve up to 2.44x speedup on a T4 GPU. Furthermore, Q-ASR exhibits a large compression rate of more than 4x with small WER degradation.
We propose an algorithm that is capable of synthesizing high quality target speakers singing voice given only their normal speech samples. The proposed algorithm first integrate speech and singing synthesis into a unified framework, and learns universal speaker embeddings that are shareable between speech and singing synthesis tasks. Specifically, the speaker embeddings learned from normal speech via the speech synthesis objective are shared with those learned from singing samples via the singing synthesis objective in the unified training framework. This makes the learned speaker embedding a transferable representation for both speaking and singing. We evaluate the proposed algorithm on singing voice conversion task where the content of original singing is covered with the timbre of another speakers voice learned purely from their normal speech samples. Our experiments indicate that the proposed algorithm generates high-quality singing voices that sound highly similar to target speakers voice given only his or her normal speech samples. We believe that proposed algorithm will open up new opportunities for singing synthesis and conversion for broader users and applications.
This paper reports on the semi-supervised development of acoustic and language models for under-resourced, code-switched speech in five South African languages. Two approaches are considered. The first constructs four separate bilingual automatic speech recognisers (ASRs) corresponding to four different language pairs between which speakers switch frequently. The second uses a single, unified, five-lingual ASR system that represents all the languages (English, isiZulu, isiXhosa, Setswana and Sesotho). We evaluate the effectiveness of these two approaches when used to add additional data to our extremely sparse training sets. Results indicate that batch-wise semi-supervised training yields better results than a non-batch-wise approach. Furthermore, while the separate bilingual systems achieved better recognition performance than the unified system, they benefited more from pseudo-labels generated by the five-lingual system than from those generated by the bilingual systems.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
