TheanoLM - An extensible toolkit for neural network language modeling

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Standard

TheanoLM - An extensible toolkit for neural network language modeling. / Enarvi, Seppo; Kurimo, Mikko.

Proceedings of the 17th Annual Conference of the International Speech Communication Association (INTERSPEECH): San Francisco, USA, Sept. 8-12. ISCA, 2016. p. 3052-3056 (Proceedings of the Annual Conference of the International Speech Communication Association).

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Harvard

Enarvi, S & Kurimo, M 2016, TheanoLM - An extensible toolkit for neural network language modeling. in Proceedings of the 17th Annual Conference of the International Speech Communication Association (INTERSPEECH): San Francisco, USA, Sept. 8-12. Proceedings of the Annual Conference of the International Speech Communication Association, ISCA, pp. 3052-3056, Interspeech, San Francisco, United States, 08/09/2016. https://doi.org/10.21437/Interspeech.2016-618

APA

Enarvi, S., & Kurimo, M. (2016). TheanoLM - An extensible toolkit for neural network language modeling. In Proceedings of the 17th Annual Conference of the International Speech Communication Association (INTERSPEECH): San Francisco, USA, Sept. 8-12 (pp. 3052-3056). (Proceedings of the Annual Conference of the International Speech Communication Association). ISCA. https://doi.org/10.21437/Interspeech.2016-618

Vancouver

Enarvi S, Kurimo M. TheanoLM - An extensible toolkit for neural network language modeling. In Proceedings of the 17th Annual Conference of the International Speech Communication Association (INTERSPEECH): San Francisco, USA, Sept. 8-12. ISCA. 2016. p. 3052-3056. (Proceedings of the Annual Conference of the International Speech Communication Association). https://doi.org/10.21437/Interspeech.2016-618

Author

Enarvi, Seppo ; Kurimo, Mikko. / TheanoLM - An extensible toolkit for neural network language modeling. Proceedings of the 17th Annual Conference of the International Speech Communication Association (INTERSPEECH): San Francisco, USA, Sept. 8-12. ISCA, 2016. pp. 3052-3056 (Proceedings of the Annual Conference of the International Speech Communication Association).

Bibtex - Download

@inproceedings{26e5e089cf4048cf91a39b3385f2108c,
title = "TheanoLM - An extensible toolkit for neural network language modeling",
abstract = "We present a new tool for training neural network language models (NNLMs), scoring sentences, and generating text. The tool has been written using Python library Theano, which allows researcher to easily extend it and tune any aspect of the training process. Regardless of the flexibility, Theano is able to generate extremely fast native code that can utilize a GPU or multiple CPU cores in order to parallelize the heavy numerical computations. The tool has been evaluated in difficult Finnish and English conversational speech recognition tasks, and significant improvement was obtained over our best back-off n-gram models. The results that we obtained in the Finnish task were compared to those from existing RNNLM and RWTHLM toolkits, and found to be as good or better, while training times were an order of magnitude shorter.",
keywords = "Artificial neural networks, Automatic speech recognition, Conversational language, Language modeling",
author = "Seppo Enarvi and Mikko Kurimo",
year = "2016",
doi = "10.21437/Interspeech.2016-618",
language = "English",
series = "Proceedings of the Annual Conference of the International Speech Communication Association",
publisher = "ISCA",
pages = "3052--3056",
booktitle = "Proceedings of the 17th Annual Conference of the International Speech Communication Association (INTERSPEECH)",

}

RIS - Download

TY - GEN

T1 - TheanoLM - An extensible toolkit for neural network language modeling

AU - Enarvi, Seppo

AU - Kurimo, Mikko

PY - 2016

Y1 - 2016

N2 - We present a new tool for training neural network language models (NNLMs), scoring sentences, and generating text. The tool has been written using Python library Theano, which allows researcher to easily extend it and tune any aspect of the training process. Regardless of the flexibility, Theano is able to generate extremely fast native code that can utilize a GPU or multiple CPU cores in order to parallelize the heavy numerical computations. The tool has been evaluated in difficult Finnish and English conversational speech recognition tasks, and significant improvement was obtained over our best back-off n-gram models. The results that we obtained in the Finnish task were compared to those from existing RNNLM and RWTHLM toolkits, and found to be as good or better, while training times were an order of magnitude shorter.

AB - We present a new tool for training neural network language models (NNLMs), scoring sentences, and generating text. The tool has been written using Python library Theano, which allows researcher to easily extend it and tune any aspect of the training process. Regardless of the flexibility, Theano is able to generate extremely fast native code that can utilize a GPU or multiple CPU cores in order to parallelize the heavy numerical computations. The tool has been evaluated in difficult Finnish and English conversational speech recognition tasks, and significant improvement was obtained over our best back-off n-gram models. The results that we obtained in the Finnish task were compared to those from existing RNNLM and RWTHLM toolkits, and found to be as good or better, while training times were an order of magnitude shorter.

KW - Artificial neural networks

KW - Automatic speech recognition

KW - Conversational language

KW - Language modeling

UR - http://www.scopus.com/inward/record.url?scp=84994235652&partnerID=8YFLogxK

U2 - 10.21437/Interspeech.2016-618

DO - 10.21437/Interspeech.2016-618

M3 - Conference contribution

T3 - Proceedings of the Annual Conference of the International Speech Communication Association

SP - 3052

EP - 3056

BT - Proceedings of the 17th Annual Conference of the International Speech Communication Association (INTERSPEECH)

PB - ISCA

ER -

ID: 9697062