English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Vocoders are models capable of transforming a low-dimensional spectralrepresentation of an audio signal, typically the mel spectrogram, to awaveform. Modern speech generation pipelines use a vocoder as their finalcomponent. Recent vocoder models developed for speech achieve a high degree ofrealism, such that it is natural to wonder how they would perform on musicsignals. Compared to speech, the heterogeneity and structure of the musicalsound texture offers new challenges. In this work we focus on one specificartifact that some vocoder models designed for speech tend to exhibit whenapplied to music: the perceived instability of pitch when synthesizingsustained notes. We argue that the characteristic sound of this artifact is dueto the lack of horizontal phase coherence, which is often the result of using atime-domain target space with a model that is invariant to time-shifts, such asa convolutional neural network. We propose a new vocoder model that isspecifically designed for music. Key to improving the pitch stability is thechoice of a shift-invariant target space that consists of the magnitudespectrum and the phase gradient. We discuss the reasons that inspired us tore-formulate the vocoder task, outline a working example, and evaluate it onmusical signals. Our method results in 60% and 10% improved reconstruction ofsustained notes and chords with respect to existing models, using a novelharmonic error metric.

Mel Spectrogram Inversion with Stable Pitch