Skip to content

Coding

Angefangen mit kleinen eigenen Games, Homepages und Experimenten mit Microcontrollern bin ich schließlich zum Informatikstudium gekommen. Es fasziniert mich Visionen in Apps, Websites und Anwendungen umzusetzen und diese dann auf dem eigenen Gerät verwenden zu können. Darüber hinaus durfte ich besonders im Rahmen meiner Bachelorarbeit erleben, wie wichtig und hilfreich die Informatik auch in anderen Bereichen zur Optimierung von Workflows und Prozessen sein kann. In meinem Masterstudium habe ich meinen Fokus auf die Entwicklung von KI-Lösungen gerichtet und durfte einmal mehr die zahlreichen Möglichkeiten entdecken, die digitale Welt von morgen mitzugestalten.

High resolution film emulation with AI generated 3D LUTs

Exploring AI driven 3D LUT generation for high resolution image editing based on multiple CNN based architectures int he context of film emulation.

Published: at 09:20 AM

Scaling an AI Transcription Model as a Service

Ton-Texter is a Software as a service solution to transcribe audio and video files and get different transcript file formats with filmmakers as our target audience. In this blog post, we will explore how we have improved its scalability to handle hig...

Published: at 05:20 AM

The current state of generative AI in the context of Cyber Security

This paper was written as part of the lecture “Secure Systems“ at Stuttgart Media University. It is an analysis of the current state of generative Artificial Intelligence (AI) in the context of cybersecurity.

Published: at 03:20 AM

Many components - one central place for logs and metrics - how Ton-Texter uses AWS Cloudwatch

Ton-Texter consists of many different components. To get an overview of whats happening it is important to have one central place for logs and all important metrics. Therefore, we use AWS Cloudwatch.

Published: at 09:59 PM

40% faster - drastically improved quality - another step up in Ton-Texters transcription

With the release of the whisper-large-v3-turbo model we managed to improve Ton-Texters transcription quality and efficiency once more. The real-time factor of transcription is increased from 8x to 11x, which means transcription is around 40% faster.

Published: at 09:59 PM

Deutsch -> Schwäbisch | Training Googles T5

Experiment to fine-tune the pretrained Text-to-Text Transfer Transformer, published by Google, to translate sentences from regular german into schwäbisch.

Published: at 05:20 AM

Using Wordpress as a headless CMS for a React page

In the first implementation of the homepage for Jari Claes I used the headless CMS Strapi. Even though the user experience was great, we decided to change the architecture, because this did come with high hosting cost. Therefore I reimplemented the b...

Published: at 09:17 PM

WebAssembly - a web performance revolution?

An evaluation of the current state of WebAssemblys performance, potential use cases, usage and remaining limitations. A paper I produced as part of the lecture "Entwicklung von Rich Media Systemen" at HDM Stuttgart.

Published: at 01:12 PM

Event processing and video upload in a ultra large scale video platform

This paper was written as part of the lecture "Ultra Large Scale Systems" at Stuttgart Media University. In the lecture, we decided to take a closer look at video platforms such as YouTube and examined several sections of its large-scale architecture...

Published: at 11:46 AM

How we improved our transcription performance by over 60%

By effiency improvements in our code and by using Faster-Whisper we improved our transcription performance significantly

Published: at 04:59 PM

Is Ton-Texter really competitive? - Yes, it is!

Ton-Texter offers state of the art transcription, whilst being considerably cheaper than its competitors

Published: at 04:58 PM

How we built our own 'serverless' GPU cloud infrastructure

Serverless Lambda functions have been known for a long time, but unfortunately there are little to no serverless GPU offerings on the market. Thats why we built our own solution with Lambda and EC2

Published: at 04:57 PM

The backbone of Ton-Texter - Transcription based on OpenAI Whisper / Pyannote

How we managed to get state of the art transcription and speaker diarization performance for our service

Published: at 04:58 PM

Using CUDA with an AMI on AWS EC2

CUDA significantly boosts AI performance on our EC2

Published: at 04:57 PM