Multi-Task Audio Source Separation (MTASS) and SpeechNAS: AutoML‑Driven Large‑Scale Speaker Recognition

This article presents two ASRU‑2021 accepted works from Kuaishou: MTASS, a multi‑task audio source separation framework that jointly separates speech, music and noise, and SpeechNAS, an AutoML‑based neural architecture search method that achieves state‑of‑the‑art speaker recognition performance with significantly fewer parameters.

AutoMLMTASSNeural Architecture Search

0 likes · 14 min read

Multi-Task Audio Source Separation (MTASS) and SpeechNAS: AutoML‑Driven Large‑Scale Speaker Recognition

58 Tech

Oct 9, 2020 · Artificial Intelligence

Speaker Role Recognition in an Intelligent Voice Analysis Platform

This article describes a speaker role recognition system for a voice analysis platform, detailing a gender‑based pre‑filter, keyword‑matching and TextCNN‑based text classification, and single‑sentence correction methods that together improve role assignment accuracy by about 6% over baseline third‑party solutions.

AINLPTextCNN

0 likes · 12 min read

Speaker Role Recognition in an Intelligent Voice Analysis Platform

DataFunTalk

Mar 10, 2020 · Artificial Intelligence

Interspeech 2019 Highlights: End‑to‑End Speech AI Technologies and Key Paper Summaries

The article reviews Interspeech 2019, summarizing major trends and representative papers in end‑to‑end speech recognition, synthesis, natural language understanding, speaker recognition, and speech translation, while also highlighting best student papers and providing resources for further study.

AIInterspeech 2019Speech synthesis

0 likes · 24 min read

Interspeech 2019 Highlights: End‑to‑End Speech AI Technologies and Key Paper Summaries