Benjamin Milde

Name:	Benjamin Milde
Position:	PhD Student
Email:	milde ä~t informatik.uni-hamburg.de
Phone:	+49 40 42883 2387
Fax:	+49 40 42883 2345
Office:	F-430
Address:	Informatikum Vogt-Kölln-Straße 30 22527 Hamburg

Hello, I am Benjamin! I am currently a doctoral student at the language technology group, under the supervision of Prof. Chris Biemann. I am interested in representation learning in speech signals, self-supervised machine learning, neural networks, automatic speech recognition (ASR) and natural language processing applications that deal with speech input.

Projects

Dialog+ / Ambient Search: A real-time document recommendation system for speech streams (git)

Open source German speech recognition models (git)

Unspeech: Unsupervised training of speech context embeddings (git, website)

Sparsespeech: Unsupervised symbolic self-labeling of speech data (git)

Subtitle2Go: Automatic subtitling of lecture videos (git).

Publications

Open Source Automatic Lecture Subtitling (pdf, code)

Benjamin Milde, Robert Geislinger, Irina Lindt, Timo Baumann

In: Proceedings of ESSV 2021

Improving Unsupervised Sparsespeech Acoustic Models with Categorical Reparameterization (pdf, code)

Benjamin Milde and Chris Biemann

In: Proceedings of Interspeech 2020, Virtual Shanghai, China

SparseSpeech: Unsupervised Acoustic Unit Discovery with Memory-Augmented Sequence Autoencoders (pdf, code)

Benjamin Milde and Chris Biemann

In: Proceedings of Interspeech 2019, Graz, Austria

Open Source Automatic Speech Recognition for German (pdf)

Benjamin Milde and Arne Köhn

In: Proceedings of 13th ITG Conference on Speech Communication 2018, Oldenburg

Unspeech: Unsupervised Speech Context Embeddings. (pdf, website, code)

Benjamin Milde and Chris Biemann
In: Proceedings of Interspeech 2018, Hyderabad, India

Multitask Sequence-to-sequence Models for Grapheme-to-phoneme Conversion

Benjamin Milde, Christoph Andreas Schmidt, Joachim Köhler
In: Proceedings of Interspeech 2017, Stockholm, Sweden

Ambient Search: A Document Retrieval System for Speech Streams

Benjamin Milde, Jonas Wacker, Stefan Radomski, Max Mühlhäuser, Chris Biemann
In: Proceedings of COLING 2016, December 2016.

Demonstrating Ambient Search: Implicit Document Retrieval for Speech Streams

Benjamin Milde, Jonas Wacker, Stefan Radomski, Max Mühlhäuser, Chris Biemann
In: Proceedings of COLING 2016 (System demonstrations), December 2016.

Classification of Speaker Intoxication Using a Bidirectional Recurrent Neural Network

Kim Berninger, Jannis Hoppe, Benjamin Milde

In: Proceedings of Text, Speech and Dialogue (TSD), 2016

NLU vs. Dialog Management: To Whom am I Speaking?

Dirk Schnelle-Walka, Stefan Radomski, Benjamin Milde, Chris Biemann, Max Mühlhäuser
In: Joint Workshop on Smart Connected and Wearable Things (SCWT'2016), co-located with IUI, March 2016.

Open-Source German Distant Speech Recognition: Corpus and Acoustic Model

Stephan Radeck-Arneth, Benjamin Milde, Arvid Lange, Evandro Gouvea, Stefan Radomski, Max Mühlhäuser, Chris Biemann
In: Pavel Král, Václav Matoušek: Text, Speech and Dialogue. Proceedings of the 18th International Conference TSD2015, Plzen, Czech Republic, p. 480-488, Springer International Publishing, September 2015.

Using Representation Learning and Out-of-domain Data for a Paralinguistic Speech Task

Benjamin Milde, Chris Biemann
In: Proceedings of the INTERSPEECH 2015, Dresden, Germany.