Build Conversational Voice AI App
Publicada el 2026-07-24
Descripción de la oferta
I’m ready to move from concept to prototype on a voice-first mobile application that will run natively on both iOS and Android. The core interaction is simple to describe yet technically demanding: a user presses the mic, speaks, their speech is converted to text with minimal latency, that text is sent straight to an LLM, and the model’s response is returned immediately inside the app. The entire flow must feel seamless and natural for a general-public audience who expect consumer-grade polish. Here is the flow I need implemented: • Microphone capture with push-to-talk (or similar) that works reliably across current iOS and Android versions. • On-device or cloud speech-to-text with high accuracy; if you have experience with Whisper, Google Speech, or Apple’s Speech framework, please highlight it. • Secure, efficient hand-off of the transcribed text to an LLM endpoint (OpenAI, Claude, or another you recommend) and return of the model’s answer. • A clean, lightweight UI that shows the running transcript and the LLM’s response in real time. Quality targets • End-to-end round-trip (speak → reply on screen) under three seconds on a decent connection. • Error handling that guides the user gracefully when connectivity drops or speech isn’t recognised. • Codebase structured for future expansion (e.g., adding text-to-speech or user accounts later). Deliverables 1. Fully functioning MVP for both iOS (Swift / SwiftUI preferred) and Android (Kotlin / Jetpack Compose preferred). 2. Clear setup notes and build scripts so I can run the project locally. 3. A short video demo and testflight / APK links for initial user testing. If you have shipped voice-driven or conversational AI apps before, or have benchmark data around speech-to-text latency and LLM throughput, that’s exactly the expertise I’m after. Let’s create an experience that feels like talking to the future.
Skills
Fuente original: freelancer