ON 25 JUNE, jazz singer and high-school music teacher Daniel Hutchins broke into a ballad at the San Antonio Convention Center in Texas, accompanied by music student Jacob Mann on piano. It was a perfectly synchronised performance 鈥� but the pair were 2000 kilometres apart. Mann played his contribution online from the University of Southern California in Los Angeles.
Their interaction was made possible by a real-time technology that looks set to change the face of musical composition, rehearsal and live performance. Not being in the same room, city or even country need no longer be a barrier to musical collaboration.
The network that Hutchins and Mann used is an Italian-developed system called 鈥� and it provides such pin-sharp audio precision that it can even be used for live music events in which some participants are only present online. In a demo in October last year, for instance, a violinist on stage in Philadelphia, Pennsylvania, used LOLA to perform a tricky duet with a cellist 1300 kilometres away in DeKalb, Illinois 鈥� to . The audience of 600 people didn鈥檛 notice any delay in the remote cellist鈥檚 playing.
Advertisement
LOLA works over the fast fibre-optic networks running between universities 鈥� such as Janet in the UK and Internet2 in the US. But online musical collaboration is also happening on the regular internet. A service called , for instance, allows bands to thrash out new songs and rehearse online. 鈥淚t gives us the ability to write spontaneously when the inspiration hits, regardless of our locations,鈥� says Dustin Bogue, a country musician from Nashville, Tennessee. Indeed, he and songwriting partner Andy Brasher say they wrote their latest song, almost entirely over the eJamming network.
What鈥檚 behind the trend is the development of ways to minimise the time delay, or latency, that online users experience, and which can play havoc with conversations, let alone precise musical timing. This latency is seen even on fast research networks and cutting-edge ventures like Google Fiber. What must be achieved, says Gill Davies, a researcher in distributed music at Edinburgh Napier University, UK, is a latency generally below 30 milliseconds 鈥� otherwise the human ear notices a delay.
Alan Glueckman, , says the site allows live audio streaming between up to four locations 鈥� so a drummer, guitarist, bassist and singer can be in separate places. It does this by eschewing the server-mediated way the web works, instead establishing a peer-to-peer network between subscribers. 鈥淭his cuts the travel time for audio packets to a minimum 鈥� as there鈥檚 no travel time to a central server and back again to each of the four locations,鈥� says Glueckman. The network also uses an internet protocol called , which requires no acknowledgement of data receipt before sending the next packet of data 鈥� slashing transmission time.
The upshot? eJamming members in the US get up to 25 ms latency if they are in the same city, 30 ms in the same state and 45 ms coast-to-coast. Even those playing with people in other countries, who experience longer latencies, appreciate the service. 鈥淛amming in real time with someone on the other side of the world is mind-boggling,鈥� says Ron Bull, a user in Surrey, British Columbia, Canada. And Mike Reid in San Francisco says: 鈥淚t鈥檚 the closest thing to being in the same studio.鈥�
Glueckman is now raising funding to upgrade to a six-location version of eJamming that will also have a video channel 鈥� allowing users to see each other for the first time. Video is already a key feature of LOLA. 鈥淰ideo is vital to classical performers,鈥� says Matt Parkin, a recording engineer at London鈥檚 Royal College of Music. 鈥淎 nod of the head or the lift of an eyebrow can mean 鈥榮tart now鈥�, 鈥榣et鈥檚 slow down鈥� or 鈥榣et鈥檚 finish this note together'鈥�.
鈥� a shortening of 鈥渓ow latency鈥� 鈥� runs on a regular PC and is free for noncommercial users. It was developed by a team led by Claudio Allocchio at the Italian Academic and Research Network in Rome, which runs an intra-university gigabit network known as GARR that is Italy鈥檚 equivalent of Internet2 and Janet.
鈥淚t鈥檚 like very fast videoconferencing with no audio compression or video processing involved to slow things down,鈥� says Parkin. The result is that encoding and decoding the sound only takes 5 ms. And as long as the participants are within 3000 kilometres of each other the overall audio latency will be acceptable to the ear.
The video runs at 60 frames per second. 鈥淭hat鈥檚 to capture important things like the bowing action of a violinist or cellist accurately 鈥� you don鈥檛 want any bowing blur when you鈥檙e watching for a cue,鈥� says Davies.
In a recent LOLA trial, a virtual clarinet quartet was assembled 鈥� with two players at the Royal College of Music in London and a pair in Edinburgh. The latency was just 7 ms, says Parkin, but what surprised him most was the rapid user acceptance. 鈥淚t was all a bit strange at first but they soon settled into it 鈥� and quickly the four of them started chatting and joking like musicians at any normal rehearsal.鈥�
Keeping time, online
Seeing the conductor鈥檚 baton move a fraction too late can leave musicians playing through a live link lagging behind an orchestra.
It can even lead to the entire orchestra slowing its tempo to compensate.
An animation that predicts where the baton will be in a few beats鈥� time could make such performances as tight as a concert that has everyone in the same room, says Donald Dansereau at the University of Sydney, Australia.
His team has written a machine learning algorithm that predicts future baton motion from recent trajectories and uses it to drive an animated baton (Computer Music Journal, ).
This article appeared in print under the headline 鈥淎 long-distance love song鈥�
