# Fast Spectrogram Inversion using Multi-head Convolutional Neural Networks

arXiv: https://arxiv.org/abs/1808.06719 · HTML: https://mlsanity.com/paper/1808.06719/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Aug 2018.

## Built on by (most cited first, 22 of 22)

- [A Survey on Neural Speech Synthesis](https://mlsanity.com/paper/2106.15561/) (2021)
- [Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safety](https://mlsanity.com/paper/2104.14235/) (2021)
- [Universal Neural Vocoding with Parallel WaveNet](https://mlsanity.com/paper/2102.01106/) (2021)
- [On loss functions and evaluation metrics for music source separation](https://mlsanity.com/paper/2202.07968/) (2022)
- [Catch-A-Waveform: Learning to Generate Audio from a Single Short Example](https://mlsanity.com/paper/2106.06426/) (2021)
- [Audio representations for deep learning in sound synthesis: A review](https://mlsanity.com/paper/2201.02490/) (2022)
- [RFWave: Multi-band Rectified Flow for Audio Waveform Reconstruction](https://mlsanity.com/paper/2403.05010/) (2024)
- [NeRAF: 3D Scene Infused Neural Radiance and Acoustic Fields](https://mlsanity.com/paper/2405.18213/) (2024)
- [Unsupervised Harmonic Parameter Estimation Using Differentiable DSP and Spectral Optimal Transport](https://mlsanity.com/paper/2312.14507/) (2023)
- [Msanii: High Fidelity Music Synthesis on a Shoestring Budget](https://mlsanity.com/paper/2301.06468/) (2023)
- [Synthesizer Sound Matching Using Audio Spectrogram Transformers](https://mlsanity.com/paper/2407.16643/) (2024)
- [Multi-Loss Convolutional Network with Time-Frequency Attention for Speech Enhancement](https://mlsanity.com/paper/2306.08956/) (2023)
- [ROSE: A Recognition-Oriented Speech Enhancement Framework in Air Traffic Control Using Multi-Objective Learning](https://mlsanity.com/paper/2312.06118/) (2023)
- [EcoFlow: Efficient Convolutional Dataflows for Low-Power Neural Network Accelerators](https://mlsanity.com/paper/2202.02310/) (2022)
- [RaD-Net 2: A causal two-stage repairing and denoising speech enhancement network with knowledge distillation and complex axial self-attention](https://mlsanity.com/paper/2406.07498/) (2024)
- [Sines, Transient, Noise Neural Modeling of Piano Notes](https://mlsanity.com/paper/2409.06513/) (2024)
- [Miipher-2: A Universal Speech Restoration Model for Million-Hour Scale Data Restoration](https://mlsanity.com/paper/2505.04457/) (2025)
- [Representation of perceived prosodic similarity of conversational feedback](https://mlsanity.com/paper/2505.13268/) (2025)
- [SPEAR: Receiver-to-Receiver Acoustic Neural Warping Field](https://mlsanity.com/paper/2406.11006/) (2024)
- [Frequency-Aware Masked Autoencoders for Human Activity Recognition using Accelerometers](https://mlsanity.com/paper/2502.17477/) (2025)
- [Efficient Neural and Numerical Methods for High-Quality Online Speech Spectrogram Inversion via Gradient Theorem](https://mlsanity.com/paper/2505.24498/) (2025)
- [Challenges of Multi-Speaker Extraction for Real Conversational Speech Enhancement](https://mlsanity.com/paper/2609.25948/) (2026)

## Abstract (arXiv)

We propose the multi-head convolutional neural network (MCNN) architecture for waveform synthesis from spectrograms. Nonlinear interpolation in MCNN is employed with transposed convolution layers in parallel heads. MCNN achieves more than an order of magnitude higher compute intensity than commonly-used iterative algorithms like Griffin-Lim, yielding efficient utilization for modern multi-core processors, and very fast (more than 300x real-time) waveform synthesis. For training of MCNN, we use a large-scale speech recognition dataset and losses defined on waveforms that are related to perceptual audio quality. We demonstrate that MCNN constitutes a very promising approach for high-quality speech synthesis, without any iterative algorithms or autoregression in computations.