LINEAiGE
AI HISTORY

WaveNet: A generative model for raw audio

Google DeepMind publicly presented WaveNet on September 8, 2016 as a deep generative model of raw audio waveforms. The associated research described WaveNet as a probabilistic autoregressive neural network that generates raw audio sample by sample and demonstrated applications including text-to-speech and music generation.

8 September 2016AI HISTORYVERIFIED

Historical classification

This verified record belongs to LINEAiGE's artificial-intelligence history layer beginning in 1956.

Why it matters

A documented generative-audio research milestone for direct modeling of raw audio waveforms with an autoregressive neural network.

People, organizations & technologies

PEOPLE
Aäron van den Oord, Sander Dieleman, Heiga Zen, Karen Simonyan, Oriol Vinyals, Alex Graves, Nal Kalchbrenner, Andrew Senior, Koray Kavukcuoglu

ORGANIZATIONS
Google DeepMind

TECHNOLOGIES
WaveNet, Autoregressive generative model, Raw audio generation, Text-to-speech

Evidence & verification

STATE
verified

CONFIDENCE
high

LAST VERIFIED
2026-09-10T05:06:56+08:00

Google DeepMind's first-party research post establishes September 8, 2016 as the public research-introduction date and directly supports the raw-audio generative-model and speech/music claims. The original WaveNet arXiv paper, first submitted September 12, 2016, independently supports the authorship, probabilistic autoregressive formulation, raw-waveform generation, text-to-speech, and music-generation claims. The event date is therefore attached specifically to the DeepMind public research introduction, not asserted as a software-release, API, deployment, or model-training completion date. No geography, ancestry, downstream influence, causality, or relationships are inferred.

Sources

SOURCE 1 · PRIMARY · CLAIM-EVIDENCE

WaveNet: A generative model for raw audio

Google DeepMind

PUBLISHED
2016-09-08

SOURCE 2 · PRIMARY · CORROBORATION

WaveNet: A Generative Model for Raw Audio

arXiv

PUBLISHED
2016-09-12

Trace the chronology

Record identity

LINEAiGE ID
wavenet-2016