
ETM-Network is a suite of on-device artificial intelligence applications and research initiatives developed by Mensch, Nina! UG. The platform focuses on deploying state-of-the-art AI models directly on user devices, emphasizing privacy by ensuring that data processing and inference occur locally without information leaving the device. Its offerings span audio processing, speech technologies, fitness tracking, and research into adaptive AI architectures.
Among its featured applications, NeuralAudio provides real-time audio denoising powered by edge AI, allowing users to remove background noise from audio locally on their phones. Neural-TTS delivers text-to-speech capabilities, generating human-like speech in near real-time with four selectable English voices, again processed entirely on-device. Neural-STT offers speech-to-text transcription using a Conformer architecture model, supporting both live recording and file import, with all transcription handled privately on the user's device. NeuralVoice is an offline audiobook reader that transforms EPUB and PDF files into natural-sounding audio using a diffusion model, maintaining user privacy through on-device inference.
The platform also introduces NeuralFit, an upcoming fitness application that uses pose estimation and computer vision to track workout form and promote digital wellbeing by locking distracting apps until workouts are completed. In addition to these user-facing tools, ETM-Network is engaged in research on adaptive AI systems through its Chimera initiative. Chimera aims to develop adaptive compute units called super-neurons, which operate on event streams, maintain internal memory, and learn locally without full backpropagation. The system features dynamic spawning of specialist units, structural plasticity, and compatibility with PyTorch tensors for rapid experimentation. Early research results indicate improved adaptability and performance compared to baseline approaches.
ETM-Network's services are intended for users aged 16 and above. The platform utilizes Firebase Auth/Firestore, Crashlytics, and Google AdMob as core services, and is hosted by Netcup in Germany. The corporate structure and legal compliance details are provided, with Mensch, Nina! UG (haftungsbeschränkt) as the official publisher and Katharina Schlude as the managing director.
ETM-Network sits in PulseGate's AI & ML category. Allowing developers to integrate privacy-preserving, on-device AI for audio and speech processing without cloud dependency. It is built as an open-source project for developers building AI-powered audio, speech, and fitness apps. The product ships for the command line, and it can be self-hosted.
Behind ETM-Network is A Mensch, Nina! UG, based in Germany, and the product first shipped in 2025. Key capabilities include on-device inference, audio denoising, and speech-to-text.
Latest indexed changes and source events
Other apps tracked under the same category.