Build/deliver ultra-realistic Dutch multi-speaker voice cloning system for Mac Mini M4 (16GB)
Budget: $250 – $750 USD
Title: Build/deliver ultra-realistic Dutch multi-speaker voice cloning system for Mac Mini M4 (16GB)
I need a COMPLETE working local solution for a Mac Mini M4 (16GB) that can generate highly realistic Dutch podcast audio with multiple speakers.
Requirements:
- Runs locally on Mac Mini M4, 16GB RAM
- Dutch speech only
- Voice cloning from MAX 10 minutes of clean voice samples per speaker
- Output must sound human-level realistic, expressive, natural, and not recognizable as standard TTS
- Reference point: quality should be in the direction of projects like Parkiet, but the final solution must be more robust, cleaner, more stable, and more production-ready
- Multiple speakers in one project (separate rendering per speaker is acceptable if final workflow is simple)
- Generation speed: max 1.5x audio duration
- Clean studio-quality output: no artifacts, glitches, metallic sound, clipping, unstable loudness, or other synthetic issues
- Final audio must be normalized and production-ready
- Reliable, repeatable workflow with documentation
Deliverables:
- Fully working local installation on my Mac Mini M4 (16GB)
- All source code, scripts, configs, dependencies, and setup instructions
- Simple end-to-end workflow from text + voice samples to final podcast audio
- Test project proving the requirements above
- Full rights to use the delivered work commercially
Mandatory acceptance criteria:
1. Must run successfully on my Mac Mini M4 16GB
2. Must pass my blind listening test for realism/naturalness
3. Must meet the max 1.5x generation-time limit
4. Must produce clean artifact-free output
5. Must be fully reproducible from the delivered instructions/files
Payment terms:
- Fixed price only
- Escrow only
- No upfront payment
- Final payment only after full delivery and successful acceptance test on my machine
- If it does not meet ALL requirements, the project is not accepted
Important:
- Do NOT bid with standard TTS, basic cloning, or “similar quality”
- Bid only if you already have proven experience with high-end voice cloning / expressive speech synthesis
- In your bid, include: relevant examples, exact approach, expected performance on Mac Mini M4 16GB, and what is custom-built vs existing open-source
- Only legal/authorized voice cloning work
I need a COMPLETE working local solution for a Mac Mini M4 (16GB) that can generate highly realistic Dutch podcast audio with multiple speakers.
Requirements:
- Runs locally on Mac Mini M4, 16GB RAM
- Dutch speech only
- Voice cloning from MAX 10 minutes of clean voice samples per speaker
- Output must sound human-level realistic, expressive, natural, and not recognizable as standard TTS
- Reference point: quality should be in the direction of projects like Parkiet, but the final solution must be more robust, cleaner, more stable, and more production-ready
- Multiple speakers in one project (separate rendering per speaker is acceptable if final workflow is simple)
- Generation speed: max 1.5x audio duration
- Clean studio-quality output: no artifacts, glitches, metallic sound, clipping, unstable loudness, or other synthetic issues
- Final audio must be normalized and production-ready
- Reliable, repeatable workflow with documentation
Deliverables:
- Fully working local installation on my Mac Mini M4 (16GB)
- All source code, scripts, configs, dependencies, and setup instructions
- Simple end-to-end workflow from text + voice samples to final podcast audio
- Test project proving the requirements above
- Full rights to use the delivered work commercially
Mandatory acceptance criteria:
1. Must run successfully on my Mac Mini M4 16GB
2. Must pass my blind listening test for realism/naturalness
3. Must meet the max 1.5x generation-time limit
4. Must produce clean artifact-free output
5. Must be fully reproducible from the delivered instructions/files
Payment terms:
- Fixed price only
- Escrow only
- No upfront payment
- Final payment only after full delivery and successful acceptance test on my machine
- If it does not meet ALL requirements, the project is not accepted
Important:
- Do NOT bid with standard TTS, basic cloning, or “similar quality”
- Bid only if you already have proven experience with high-end voice cloning / expressive speech synthesis
- In your bid, include: relevant examples, exact approach, expected performance on Mac Mini M4 16GB, and what is custom-built vs existing open-source
- Only legal/authorized voice cloning work