WhisppWAVE
for Mobile
Real-time Voice Reconstruction AI on Android. Turn a whisper into a clear, natural voice, or anonymize your voice, on-device.
A clear, natural voice, reconstructed in real time.
WhisppWAVE for Mobile reconstructs speech as it happens, so a whisper or a voice comes through clear and natural on the call. It works in any language and adds no perceptible delay.
What you're integrating
Android
Android phones and tablets. Mobile chipsets, leveraging NPU support.
Below 120 ms
Total AI latency under 120 ms for live calls.
On-device
Runs on the device CPU or NPU.
Real-time voice reconstruction, built into your existing stack.
WhisppWAVE for Mobile is a native Android SDK with drop-in integration into your existing audio pipeline. It runs on the device CPU or NPU, so there is no cloud dependency and no per-call server cost. One SDK with established audio licensing.
std::vector<float> WhisppInference::processFrame(const std::vector<float>& frame) {
std::vector<Tensor> inputs = {
makeTensor({1, frame.size()}, frame),
makeTensor(_voiceShape, _voiceProfile),
makeTensor({1, _streamState.size()}, _streamState)
};
auto outputs = _session.run(inputs);
_streamState = outputs[1].toVector<float>();
trackLatency(_modelName);
return outputs[0].toVector<float>();
}Use cases
The same real-time reconstruction technology, shaped around three different needs.
