I will build whisper speech to text, text to speech and voice ai apps


About this gig
Need accurate speech to text, natural text to speech or a voice feature in your product? I build custom voice AI in Python, not just no-code setups.
I'm Kethaka, a backend and machine learning engineer. At a New Zealand company I trained and tested voice models with PyTorch, TensorFlow and Whisper for speech to text, text to speech, voice cloning and speaker recognition, and built the FastAPI backends behind them.
WHAT I BUILD
Whisper transcription with timestamps and subtitles
Speech to text and text to speech APIs
Speaker recognition and speaker labels
Fine-tuning Whisper on your accents or domain words
Custom TTS voices, only with the speaker's consent
Voice features inside web and mobile apps
Python, FastAPI, PostgreSQL, Docker, cloud deploy
WHY ME
Hands-on voice ML experience in a real company
Custom code you own, no monthly tool lock-in
Clear accuracy tests on your own audio
HOW WE WORK
1. Send me a sample of your audio and your goal
2. I test it and suggest the best model and setup
3. I build, measure accuracy and tune it
4. Delivery with code, API docs and a setup guide
Send me a short audio sample and I will tell you what accuracy to expect.
Get to know Janindu Hirun
- FromSri Lanka
- Member sinceSep 2019
Languages
English, Sinhala
My Portfolio
FAQ
Which languages do you support?
Whisper supports 90+ languages. I test your language on a sample first so you know the accuracy before ordering.
Can it run on my own server?
Yes. I can run Whisper locally or on your cloud server so your audio never leaves your system.
Can you clone a voice?
Only with clear written consent from the person whose voice it is. I will not clone voices without permission.
Can you improve accuracy for my accent or industry words?
Yes. I can fine-tune Whisper on your own audio, which helps with accents and technical terms.

