Welcome back to Beyond the Patterns. Today I have the great pleasure to announce a new

colleague here at Friedrich Alexander University along Nürnberg and his name is Tobias Reichenbach.

So he leads a research group on sensory neuroengineering here at our university, did his masters in

physics at the University of Leipzig and his PhD also in physics at the Ludwig Maximilian

University in Munich. Previously he worked with the Kavli Prize winner Professor Hasbeth

at the Rockefeller University in New York and led a research group at Imperial College

London. His multi-disciplinary research combines methods from artificial intelligence with

computational neuroscience and neuroimaging to advance our understanding of the neuroprocessing

of complex natural signals with applications in medicine and technology. He has published

more than 50 peer-reviewed articles of which some have appeared in leading multi-disciplinary

journals such as Nature, Neuron and the Proceedings of the National Academy of Sciences. Dr.

Reichenbach is also a reviewing editor for eLive, a renowned journal in the life sciences.

Today we have the great pleasure to hear a presentation by him and it's entitled Decoding

the Neuroprocessing of Speech. Tobias, it's a great pleasure to have you here and the

stage is yours.

Thank you very much for the very kind introduction and of course also the invitation to speak

here also in this online form and in this recorded form. I think it's a great way to

reach a really good audience. So as I said, I work on sensory neuroengineering and just

wanted to broadly introduce the aims that we have there. So we typically think of our

brain as processing five different kinds of information if you want, us having five different

sensors, vision, touch, smell, hearing and taste. And the general aims that we have in

sensory neuroengineering is to on the one hand understand the neuro-voluntary of sensing,

then second to also aid people or to understand impairments that occur with the processing

of these different sensor information and third to also then of course rehabilitate

or help people to overcome some of the limitations that come with these impairments. Now a lot

of my work really focuses on speech and hearing and one of the main difficulties or as I said,

let's first go from the healthy person. So for us as healthy people that have normal

hearing we have an amazing ability to understand speech even in difficult listening environments.

So many everyday environments that we are in are often quite noisy. Think of car traffic

or if you're in a busy bar or restaurant and you have a lot of other background noise

around you. And despite that background noise as healthy individuals we're very well able

to nonetheless understand a particular target person and have a conversation with them.

And I would like to give you some audio demonstration to illustrate that this is really a non-trivial

problem. So I have several audio examples. I will ask you to listen to these and see

what you can understand of a particular target speech. So the first example that I want to

play you, you will hear a female voice amidst background noise. So there will be several

background speakers. Actually the audio recording will start without background noise. And then

after a few seconds that female voice will also start to say one sentence. That sentence

will start with a word footprints. So please listen for that word footprints and try to

hear if you can understand that sentence that that female person is saying.

So who heard the word footprints? I don't know if I can see. It might be difficult to

have a show of hands here. I think there's some comments. Okay, that's great. Did anybody

understand what that person was saying? I would expect not. So it's really actually,

although I said we are very good at understanding speech and background noise. That example

that I played you had a very high noise level. So actually the noise was had a higher intensity

than that target speech. But I want to give you a bit of a cue. So next I'll play you

another sentence by that same female target voice, but without background noise. So that's

of course easy to understand, but it will give us some idea of how that target voice

sounds like. A white silk jacket goes with any shoes.

All right, that's easy to understand. Now you have some idea basically, or we have some