Mittul Singh

Hello World!

I am Mittul Singh. This is my story so far!
<pic not found>


04/22: Awarded the Technology breakthrough award by Huawei Finland R&D Center

09/19: I will be presenting my recent work on Subword LMs for Automatic Speech Recognition at Interpseech 2019.


My main interest lies in intersection of speech and language technology. Currently, I am enjoying exploring deep learning methods and language theory for speech technology in low-resource scenarios.

View Publications »

Affiliations /Education

12/20: I moved to Silo AI

04/18: I moved to the Speech Recognition Group at Aalto University.

View more »

Other Interests

I love cooking, traveling, watching movies and hanging out with friends. When I am alone I like to lie in my bed and read a book until I fall asleep.

View more »


Email: mittul.singh at aalto dot fi


His name is Mittul Singh, a 30 year old guy who wants to learn. Presently, he is working as a postdoctoral researcher with Prof. Mikko Kurimo at the Speech Recognition Group, Aalto University, Finland.

He was home-born in Dhampur, India and owing to his dad’s job have so far lived in 11 of the 29 states that make India. This has given him a glimpse of diversity in India and the openness to experience the new ones around the world.

He fell in love with Mathematics during his Master‘s degree program in Mathematics and Computing at the Indian Institute of Technology Delhi (IIT Delhi). His love affair continued when in 2011 he enrolled as a Ph.D. candidate at the Spoken Language Systems Group at Saarland University, Germany , where he explored making robust and simple mathematical models for languages. This involved developing modern language models to further research in Natural Language Processing tasks like Sentence Completion, Speech Recognition, Spoken Term Detection and Event Summarisation. He has continued to pursue these interests with a special focus on low-resource data sets at his present position.

Among other interests, he likes to travel, read comics and books , try new vegetarian cuisines and write about himself in third person.

Affiliations / Education




Aku Rouhe, Astrid Van Camp, Mittul Singh, Hugo Van Hamme, Mikko Kurimo.
An Equal Data Setting for Attention-Based Encoder-Decoder and HMM/DNN Models: A Case Study in Finnish ASR
International Conference on Speech and Computer, St. Petersburg, Russia


Luca Molteni, Mittul Singh, Juho Leinonen, Katri Leino, Mikko Kurimo, Emanuele Della Valle.
Service registration chatbot: collecting and comparing dialogues from AMT workers and service’s users
Workshop on Noisy User-generated Text 2020

Hemant Kathania, Mittul Singh, Tamás Grósz, Mikko Kurimo.
Data augmentation using prosody and false starts to recognize non-native children's speech
Interspeech, 2020. Shanghai, China.

Katri Leino, Juho Leinonen, Mittul Singh, Sami Virpioja, Mikko Kurimo.
FinChat: Corpus and evaluation setup for Finnish chat conversations on everyday topics
Interspeech, 2020. Shanghai, China.

Mittul Singh, Peter Smit, Sami Virpioja, Mikko Kurimo.
Effects of Language Relatedness for Cross-lingual Transfer Learning in Character-Based Language Models
In Proceedings of the 1st Joint Workshop on Spoken Language Technologies for Under-resourced languages (SLTU) and Collaboration and Computing for Under-Resourced Languages (CCURL)


Mittul Singh, Sami Virpioja, Peter Smit, Mikko Kurimo.
Subword RNNLM Approximations for Out-Of-Vocabulary Keyword Search
In Proceedings of Interspeech, 2019. Graz, Austria.

Debjit Paul, Mittul Singh, Michael A. Hedderich, Dietrich Klakow.
Handling Noisy Labels for Robustly Learning from Self-Training Data for Low-Resource Sequence Labeling
In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Student Research Workshop.

Matthias Kleinert, Hartmut Helmke, Gerald Siol, Heiko Ehr, Dietrich Klakow, Mittul Singh, Petr Motlicek, Christian Kern, Aneta Cerna, Petr Hlousek.
Adaptation of Assistant Based Speech Recognition to new domains and its Acceptance by Air Traffic Controllers
In Proceedings of the 2nd International Conference on International Conference on Intelligent Human Systems Integration, 2019.


Matthias Kleinert, Hartmut Helmke, Heiko Ehr, Christian Kern, Dietrich Klakow, Petr Motlicek, Mittul Singh, Gerald Siol.
Building Blocks of Assistant Based Speech Recognition for Air Traffic Management Applications.
In Proceedings of the SESAR Innovation Days 2018.

Matthias Kleinert, Hartmut Helmke, Gerald Siol, Heiko Ehr, Aneta Cerna, Christian Kern, Dietrich Klakow, Petr Motlicek, Youssef Oualil, Mittul Singh, Ajay Srinivasamurthy.
Semi-supervised Adaptation of Assistant Based Speech Recognition Models for different Approach Areas.
In Proceedings of the Conference on Digital Avionics Systems Conference (DASC), 2018. (Best Paper)

Nima Nabizadeh, Mittul Singh, Dietrich Klakow.
Outperforming Neural Readers Using Broad-Context Discriminative Language Modeling on the LAMBADA Word Prediction Task.
In Proceedings of the Conference on Natural Language Processing (KONVENS), 2018. Vienna, Austria.

Ajay Srinivasamurthy, Petr Motlicek, Mittul Singh, Youssef Oualil, Matthias Kleinert, Heiko Ehr, Hartmut Helmke.
Iterative Learning of Speech Recognition Models for Air Traffic Control.
In Interspeech, 2018. Hyderabad, India.

Mittul Singh, Arunav Mishra, Youssef Oualil, Klaus Berberich, Dietrich Klakow.
Long-Span Language Models for Query-focused Unsupervised Extractive Text Summarization.
In Proceedings of the 40th European Conference on Information Retrieval (ECIR), 2018. Grenoble, France.


Mittul Singh, Youssef Oualil, Dietrich Klakow.
Approximated and domain-adapted LSTM language models for first-pass decoding in speech recognition.
Interspeech, 2017. Stockholm, Sweden.

Xiaoyu Shen, Youssef Oualil, Clayton Greenberg, Mittul Singh, Dietrich Klakow.
Estimation of gap between current language models and human performance.
Interspeech, 2017. Stockholm, Sweden.


Mittul Singh, Clayton Greenberg, Youssef Oualil, Dietrich Klakow.
Sub-Word Similarity based Search for Embeddings: Inducing Rare-Word Embeddings for Word Similarity Tasks and Language Modelling.
COLING, 2016. Osaka, Japan.

Youssef Oualil, Mittul Singh, Clayton Greenberg, Dietrich Klakow.
Long-Short Range Context Neural Networks for Language Modeling.
EMNLP, 2016. Austin, USA.

Youssef Oualil, Clayton Greenberg, Mittul Singh, Dietrich Klakow.
Sequential Recurrent Neural Networks for Language Modeling.
Interspeech, 2016. San Francisco, USA.

Mittul Singh, Clayton Greenberg, Dietrich Klakow.
The Custom Decay Language Model for Long Range Dependencies.
19th International Conference on Text, Speech and Dialogue (TSD), 2016. Born, Czech.


Benjamin Roth, Tassilo Barth, Michael Wiegand, Mittul Singh, Dietrich Klakow.
Effective Slot Filling Based on Shallow Distant Supervision Methods.
CoRR 2014.


Mittul Singh, Dietrich Klakow.
Comparing RNNS And Log-linear Interpolation Of Improved Skip-model on four BABEL Languages: Cantonese, Pashto, Tagalog, Turkish.
ICASSP 2013. Vancouver, Canada.

Benjamin Roth, Tassilo Barth, Michael Wiegand, Mittul Singh, Dietrich Klakow.
Effective Slot Filling Based on Shallow Distant Supervision Methods.
Text Analysis Conference (TAC KBP), 2013. Gaithersburg, Maryland USA.


Mittul Singh, Jayadeva, Jivitej Chadha, Puneet Ahuja, Suresh Chandra.
Reduced twin support vector regression.
Neurocomputing 2011.


Mittul Singh, Peter Smit, Sami Virpioja, Mikko Kurimo.
First-pass decoding with n-gram approximation of RNNLM: The problem of rare words.
Workshop on Machine Learning in Speech and Language Processing Workshop (MLSLP), 2018. Hyderabad, India.

Clayton Greenberg, Mittul Singh, Dietrich Klakow.
Morphological Neighbors Beat Word2Vec on the Long Tail.
1st Workshop on Deep Learning Approaches for Low-Resource Natural Language Processing (DeepLo), 2018. Melbourne, Australia.

Clayton Greenberg, Mittul Singh, Youssef Oualil, Dietrich Klakow.
Rare Word Embeddings with Sub-word Similarity Based Search: Better Representations without More Knowledge, Data, or Training.
The 2017 Israel Seminar of Computational Linguistics (ISCOL), 2017. Jerusalem, Israel.


Mittul Singh
Handling long-term dependencies and rare words in low-resource language modelling.
Ph.D. Thesis, Universität des Saarlandes, 2017.

Mittul Singh
Application of Mathematical Programming to Support Vector Machines.
Master's Thesis, IIT Delhi, 2010.

Other Interests

  • As a kid I was amazed by how my mother could make puffed indian breads (roti). It made me want to learn how to do it too. Now I can proudly say that I can do that and cook the rest of the Indian curries to my satisfaction. However, this has come with its ups and downs
  • Moving all over India and discovering a new place and an old place all over again, has made me want to travel the world even more
  • Leisure comic reading has become a passion. I read comic books while I am travelling, at home and even sometimes when cooking
  • Occasionally I write prose, poems and blogposts