How to give audio tracks “Upmix” Vintage tracks and give them new life
AudioSourceRE and Audionamix’s Xtrax Stems are the first software options available to consumers for automated decomposition. Give a song to Xtrax, for example, and the software throws out vocals, bass, drums, and “other” tracks, making for a great uplift for the soundtrack that is heard in most music in the latter period. In the end, perhaps a one-size-fits-all app will mix the actual and instant recording in its entirety; until then, it’s a track at the same time, and it’s becoming an art of its own.
What the ears can hear
On Abbey Road, James Clarke began his dismantling project in earnest around 2010. In his research, he came across a paper written in the 70s with the technique used to break video signals in component images, such as faces and backgrounds. . On paper he was reminded of his time as a master’s student in physics, working with spectrograms that show the variable frequencies of a signal over time.
Spectrograms can see signals, but the technique described on paper — called non-negative matrix factorization — was a way of processing information. If this new technique worked for video signals, it could also work for audio signals, Clark thought. “I started to study how the instruments formed a spectrogram,” he says. “I could start by saying,‘ That’s what a drummer looks like, that looks like a voice, that looks like a bass guitar. ’” A year later, he created a software that could do a convincing job of breaking. audio according to its frequencies. His first great progress can be heard 2016 remaster The Beatles’ Live in the Hollywood Bowl, the band’s only official live album. The original LP, released in 1977, is hard to hear because of the loud shouts of the people.
After trying to reduce people’s noise, Clarke finally had a “moment of serendipity”. Instead of treating the fan fan as a noise signal in the signal to be cleaned, he decided to model the fan as another tool in the mix. Identifying the crowd as his individual voice, Clarke managed to tame the Beatlemaniacs, isolating them and moving them into the background. Then that moved the four musicians sonic foreground.
Clarke was an expert in the high blending industry. He helped rescue 38 nominees for the Grammy CD Woodstock – Back to the Garden: Definitely 50th Anniversary Archive, Aiming to put together all the performances of the 1969 mega-festival. (Disclosure: I gave notes to the set online.) In one of the festival’s heaviest rains, the virtuoso sitar Ravi Shankar took to the stage. The biggest problem with the recording of the performance was not the rain, however, at Shankar’s time the producer escaped with multi-track tapes. After hearing it again in the studio, Shankar found it useful and released the fake studio At the Woodstock Festival Instead of an LP, Woodstock himself without a note. The festival’s multi-track originals were long gone, leaving future reissue producers with only a mono sound recording damaged on the concert’s soundtrack.
Using only this monophonic recording, Clarke was able to distinguish the sitar master’s instrument from the rain, the raw sonority, and the board player separated by a few feet. The result was “completely true and accurate,” with boxing co-producer Andy Zax saying the atmosphere is mixed.
“The recovery options that allow us to recover from the unrecoverable are really exciting,” Zax says. Some may see the technique as similar to coloring classic black and white films. “There’s always that tension. You want to be reconstructive and you don’t really want to impose that. That’s the challenge. “
Towards the Deep End
At the time, Clarke ended up working for the Beatles Hollywood ship the project, he and other researchers came up against a wall. Their techniques handled relatively simple models, but they could not withstand instruments with many vibrato, with subtle changes in the tones of some of the instruments and the human voice. Engineers realized they needed a new approach. “That’s what drove the depth of learning,” says Derry Fitzgerald, founder and CEO of music software company AudioSourceRE.