 Hello and welcome to a presentation for the haptics in post-2022. My name is Andreas Null, I'm the chair of Media Technology at the Technical University of Munich and I want to show you our work on automated quality assessment for compressed fiber tactile signals. So first what is quality assessment? In this area we ask the question what is the quality of a signal to a human user? And the problem is that metrics such as SNR and PSNR cannot reflect this. At the same time experiments to get this quality are very time-consuming, very inefficient. So the best would be if we have some computed metrics that we get from signal data to give us the subjective quality scores. In this scenario now we look at lossy codecs. We have an input signal going into lossy codec and a compressed signal coming out of it and then the subjective quality is defined as a similarity between the compressed signal and the input signal. To measure the similarity score we have developed a piece of software and prior work that we can use in this work to measure now data for all three state-of-the-art codecs and we get the results that you can see here as terms of score over compression ratio for all three codecs and can assess the performance between them. Now we want to introduce our perceptual metric, we call it the spectral perceptual quality index. Here we first take signal blocks and we compute the spectra of them and then we subtract the absolute threshold of perception and so we get some perceptually weighted spectra. Then these are subtracted from each other and averaged overall samples and with this we get EP which is a perceptual error measure between the different signal blocks. By mapping that error measure to a range between zero and one with XI function we can get a score in terms of percent. The results are very interesting in this case. We see that the SPQI for the blue and red codec are very close to the actual scores whereas for the green codec it's quite far off and the state-of-the-art metric so far which is the ST-SIM is the other way around it's very close for one codec and very far away for the other and this motivated us to do ViberMaf which stands for ViberTactileMotion method assessment fusion. In this case now we take the individual metrics that we have computed and we fuse them using a support vector machine into a final ViberMaf score and this ViberMaf score is the only one that can give us a score that is quite close to the actual scores for all three codecs. This is also clear with this table here we compute the Pearson correlation and MSE between the computed scores and actually measured scores and you can see that ViberMaf performs best overall for all three codecs. We've uploaded our entire code onto Github you can find all our methods on SPQI and ViberMaf there and compute your own scores on your signal data. With this I thank you very much for your attention.