01
Two-way voice translation
Support a conversation where each participant speaks their own language.
Independent builder · Pakistan · Open to collaboration
I build AI that has to work in real time.
I’m an independent developer focused on practical AI applications—especially voice, real-time pipelines, and the engineering needed to turn model capabilities into usable software.
01 / Selected work
ClearCommit
A real-time voice translation desktop app, built around a simple requirement: the conversation should keep moving while the software listens, translates, and speaks.
Python · Windows desktop · Voice AI
Speak. Understand. Keep going.
ClearCommit is a Python desktop application for two-way live voice translation. One side’s speech is recognized and translated into English captions, while the user’s microphone is translated into the other language and delivered as speech.
The project focuses on connecting speech recognition, translation, and text-to-speech into a usable live pipeline—not treating each model call as an isolated demo.
PythonCustomTkinterGroq / WhisperLLM translationDeepgram AuraElevenLabs
Technology names describe components used in the project; exact availability depends on local configuration and provider credentials. No credentials are included on this page.
FIG. 01 — Live two-way interpretation (Greek → English)Click to enlarge
FIG. 02 — Main interface: captions, controls, AI workspaceClick to enlarge
FIG. 03 — Live demo: English → Greek interpretation over Discord70 seconds
What the project is designed to do
02
Live captions and speech
Turn recognized speech into translated text and translated audio through a desktop interface.
03
Real-world audio handling
Account for issues such as noisy input, echo, recognition errors, and the timing of live turns.
02 / Honest status
What works, and what’s still being built.
The core interpreter is the established part of the project. Additional AI coaching and meeting workflows should be treated as in development until individually connected and validated in real conversations.
Core interpreter
- Python Windows desktop interface.
- Two-way live voice translation workflow using speech recognition, LLM translation, and text-to-speech components.
- Transcript display and audio-pipeline safeguards have been developed as part of the app.
- Real-world quality and latency still need to be judged on the actual machine and network used for a demo.
AI companion features
- Communication coaching for tone, phrasing, and clarity.
- Meeting intelligence for decisions, action items, and unresolved questions.
- Evidence-grounded assistant responses based on the conversation transcript.
- These features require separate provider integration and testing; this page does not claim they are production-ready.
03 / How I work
Make it useful. Test it. Keep the working parts.
I’m most interested in the unglamorous work that makes an AI feature dependable: wiring systems together, finding failure modes, measuring real behavior, and improving the product without breaking what already works.
Tools & focus areas
Python application developmentLLM integrationSpeech recognitionText-to-speechReal-time pipelinesLatency & reliabilityDebuggingTest-first iterationDesktop UX
Ways I can contribute to a team
AI integration
Connect model APIs to a real product flow, manage inputs and outputs, and keep optional AI features from breaking the core experience.
Implementation & debugging
Trace issues across components, reproduce failures, and make focused changes instead of rewriting working systems unnecessarily.
Testing under pressure
Define a small testable scope, verify behavior with realistic scenarios, and report what works, what fails, and what remains uncertain.
Product-minded engineering
Keep the demo centered on a real user problem, a coherent workflow, and evidence that the feature actually works.
04 / Contact & collaboration
Building something real?
I’m interested in collaborating with people who want to build, test, and ship a useful prototype. For a hackathon, I prefer a clear scope, frequent communication, and concrete deliverables.