1 Prompt-Based Speech-to-Voice Generation System PFE
1 Prompt-Based Speech-to-Voice Generation System PFE
Lanterns Studios•Tunisie
Speech SynthesisAudio ProcessingGame Development
Publié il y a 9 jours
Stage
⏱️4-6 mois
💼Hybride
💰Rémunéré
📅Expire dans 5 jours
Cohérence LinkedIn / CV vérifiée.
Description du poste
Project overview
Creation of an AI voice generation system that allows users to control vocal traits through text prompts.
System supports control of emotions, gender, tone, age and energy; it can optionally detect emotional cues from the script context.
Core features include prompt-based vocal style control, a voice blendspace (age, gender, tone, emotion), neutral vs emotion-following modes, and batch generation for cutscenes and dialogues.
Objectives and responsibilities
Implement prompt parsing and mapping from textual prompts to vocal-parameter controls (emotion, age, gender, tone, energy).
Design and implement a voice blendspace to interpolate voices along age/gender/tone/emotion axes and support neutral vs emotion-following output modes.
Build batch generation workflows for producing dialogue and cutscene audio at scale and integrate optional emotion detection from script context.
Integrate the generation pipeline with Unreal Engine so generated audio can be used directly in scenes and gameplay sequences.
Technical stack and tools
Work with Text-to-Speech APIs and custom audio processing components to synthesize high-quality voices.
Primary development in Python; integration with Unreal Engine required for runtime and tooling integration.
Use audio processing tools for feature extraction, post-processing, mixing and final delivery of assets.
Deliverables and evaluation
Deliver a working prototype that demonstrates prompt-based control, voice blendspace interpolation and both neutral/emotion-following modes.
Provide example batch generation scripts, Unreal Engine integration examples (demo scenes or plugins), and a short evaluation report (quality, latency, and robustness).
Include code, documentation, sample prompts, and generated audio examples for review.
Use the email subject: "Application — 1 Prompt-Based Speech-to-Voice Generation System PFE" when applying.
For questions about the project scope, include a short outline of your proposed approach and prior experience with speech synthesis or Unreal Engine integration.