Real time spectrogram inversion on mobile phone

TL;DR

This paper compares two real-time spectrogram inversion methods, streaming Griffin Lim and streaming MelGAN, analyzing their perceptual quality, latency, and resource usage on mobile devices, with insights into lookahead benefits.

Contribution

It introduces streaming versions of Griffin Lim and MelGAN for real-time spectrogram inversion on mobile phones, highlighting trade-offs and conditions for comparable quality.

Findings

Lookahead improves MelGAN perceptual quality significantly.

Streaming Griffin Lim is faster and uses less memory than MelGAN.

Both methods perform well on noisy speech without mel transformation.

Abstract

We present two methods of real time magnitude spectrogram inversion: streaming Griffin Lim(GL) and streaming MelGAN. We demonstrate the impact of looking ahead on perceptual quality of MelGAN. As little as one hop size (12.5ms) of lookahead is able to significantly improve perceptual quality in comparison to its causal version. We compare streaming GL with the streaming MelGAN and show different trade-offs in terms of perceptual quality, on-device latency, algorithmic delay, memory footprint and noise sensitivity. For fair quality assessment of the GL approach, we use input log magnitude spectrogram without mel transformation. We evaluate presented real time spectrogram inversion approaches on clean, noisy and atypical speech. We specified conditions when streaming GL has comparable quality with MelGAN: noisy audio and no mel transformation. Streaming GL is 2.4x faster than real time on the ARM CPU of a Pixel4 and it uses 4.5x times less memory than MelGAN.