AudioShake Research

AudioShake's Research is focused on isolating high-quality stems from audio and transcribing lyrics at an industry-leading level of accuracy.
AUDIOSHAKE GITHUBMEET THE TEAM

Pushing the Boundaries of Speech AI: Solving High-Fidelity Multi-Speaker Separation

AudioShake’s hi-fi multi-speaker separation isolates voices in film, TV, and AI workflows. Now available via web + API.
AudioShake Research

"Lyrics transcription for humans": revisiting AudioShake's transcription benchmark

New results from lyrics transcription benchmark showcase AudioShake's state-of-the-art transcription service
AudioShake Research

New AudioShake models double in transcription accuracy, over 5X in speed

Our new lyric transcription and alignment models nearly double accuracy, and are over 5X faster.
AudioShake Research

Stems for a wide range of guitar sounds

Latest AudioShake stems pick up different guitar styles, genres, and pedal effects with even better clarity and tone.
AudioShake Research

What makes for higher quality piano stems?

How AudioShake’s piano stems compare to open-source models, including findings from AudioLabs.
AudioShake Research

AudioShake wins Sony Demixing 2023

AudioShake's AI models earn the highest overall SDR scores in Sony challenge across music and film separation.
AudioShake Research

New benchmark for HQ transcriptions

AudioShake paper sets a new direction for high-quality lyrics transcription with a new evaluation benchmark.
AudioShake Research

New lyric transcription & alignment

LyricSync creates clean, time-stamped lyric transcriptions with high accuracy using AudioShake's vocal isolation tech.
AudioShake Research

AudioShake wins Sony Demixing Challenge

Sony ran a “Demixing Challenge” to find the world’s best song separation technology…and we won.
AudioShake Research
Research

Partners & Conferences