Research Scientist, Text-to-Speech

Oddin; Netherlands; On-site; EUR 90,000 - 120,000; Full time

Job summary

Valka, a spin-off from the Realms Group and associated with Oddin .gg, is seeking researchers to push forward real-time, co-created digital voice experiences. You will train cutting-edge TTS models, investigate architectures, and rapidly translate findings into production-ready systems.

The role requires strong ML fundamentals, Python expertise, and a passion for staying ahead of the latest in speech synthesis research. Collaborative team environment with support for deployment infrastructure.

Qualifications

• Experience with training text-to-speech or voice cloning models.
• Solid knowledge of modern ML architectures used in speech and audio.
• Proficiency in Python and ML libraries.

Responsibilities

• Research and train fast and quality SOTA TTS models for realistic voice generation.
• Experiment with architectures and data to improve quality and speed and put results to production.
• Stay up to date with research and generate ideas for improvements.
• Collaborate with a team of researchers, supported by engineering and hardware for deployment.

Skills

Text-to-speech models

Transformers

Diffusion models

GANs

Speech processing

Python

Tools

PyTorch

Hugging Face Transformers

Job description

About Valka, a visionary spin-off from the Realms Group (the parent company of Oddin .gg), is on a mission to revolutionize the way people create and experience digital content. Our team believes that content shouldn’t just be consumed; it should be co-created in real time, blurring the lines between imagination and reality. By harnessing the power of cutting-edge AI, we aim to build an interactive human-digital platform where virtual characters respond dynamically to each user’s voice, text, gestures, and more. This is your chance to join a diverse group of innovators who are driven to redefine what’s possible in generative content. Together, we’re changing the paradigm from passive viewing to active participation, unlocking new creative frontiers across gaming, entertainment, education, and beyond.

What you will be doing

• Research and train fast and quality SOTA TTS models for realistic and emotional voice generation for entertainment and education applications.
• You will be experimenting with different architectures / data to improve the quality and speed of the TTS model(s) and put the best results to production.
• Staying up to date with current research and coming up with new ideas / what to improve is very important for us!
• You will be in immediate collaboration with a team of 3 researchers specializing in TTS, and the product is supported by engineering and hardware stuff to ensure deployment

Skills you need

• Experience with training some text-to-speech / voice cloning models
• Solid knowledge of transformers, diffusion models, GANs
• Understanding of human speech and audio processing ( sampling, spectrograms, vocoders )
• Proficiency in Python and key libraries (e.g., PyTorch, Hugging Face Transformers).
• Ability to keep up to date with research, understand papers, implement approaches; strong ML fundamentals and critical thinking

Nice-to-have:

• Familiarity with modern speech synthesis models (GPT-based, flow matching... such as Vevo, StyleTTS, IndexTTS, Maskgct etc.)
• Contributions to open-source AI tools or research publications in Speech processing field
• Familiarity with AWS / similar clusters