There’s nothing quite like watching your favourite artist perform live. From the crowd’s energy to the selection of musicians for that show, a live performance rarely sounds like the record.

But how different were Glastonbury 2022 performances from their studio-recorded counterparts? And on a sonic level, what makes for a memorable performance?

We analysed some of the must-watch performances from the BBC’s Glastonbury coverage and learn that approaches to performing at the UK’s most well-known festival are extremely diverse.

Average energy is up vs recordings

Energy was the main area where we were expecting to find differences. Facing an audience of thousands, you might imagine it’s hard for artists not to dig a little deeper.

And, on average, we see an upward energy shift for Glastonbury performances over recordings. However, there was no measurable change for most tracks in our dataset. Based on the BBC’s must-see tracks, we could detect energy increased in only 38 per cent, while four per cent decreased.

The artist in our dataset that deviated the most from his recordings – and pulled the average energy up considerably – was AJ Tracey, who fielded a rock/metal backing band for the festival.

This was a massive departure from his records' minimalist hip hop and garage sound. This change increased the energy rating of his performances from medium (3/5) to very high (5/5).

Meanwhile, even though headliner Billie Eilish has a visually impressive set laden with pyrotechnics, her performance of ‘Oxytocin’ maintained a medium (3/5) rating – the same as the studio recording.

Rock tops Glasto, but hip hop reigns on record

Among the must-see tracks on BBC Music’s YouTube channel, there’s a marked difference in genre composition compared to studio recordings. The necessity for live instruments largely fuels that.

Again, AJ Tracey is one of the most extreme examples. On his records’ Ladbroke Grove’ and ‘Rain’ the AI detects hip hop, UK grime and breakbeat. But for his Glasto performance, the genres detected were rock, metal and ska.

Likewise, Africa’s biggest-selling artist Burna Boy scores hip hop and afrobeat tags for the track ‘Ye’, but genre tags are gospel and metal on the festival stage. The mix of live choir, horn section and distorted guitars certainly made it one of the most memorable of the weekend.

Changing the sound of songs for live shows isn’t the only way to go, though. Megan Thee Stallion and Kendrick Lamar used studio-recorded backing tracks on stage and supplemented with dancers instead. As expected, the AI produced similar genre tags for these performances compared to studio versions.

For catalogue managers with multiple versions of the same song, this demonstrates how easy it can be to find more unusual performances to fit specific briefs or use cases.

Angry and Exciting moods rise the Glasto ranks

One feature we can see among the must-watch tracks from Glastonbury is the higher number of angry and exciting mood tags when compared to studio recordings. We believe this can be attributed to the artists working harder to enthuse the crowd and not holding back in their vocal delivery, with more adlibs and shouting.

Tempos are remarkably consistent, bar a couple

It might seem odd for anyone who’s ever played on a stage, but tempos are remarkably similar for performances compared to their recorded counterparts. Only a handful of tracks deviated more than a couple of BPM from their recorded tempos. The track with the biggest BPM shift was Wet Leg’s ‘Chaise Longue’, which was a full 15 BPM faster than the record, which made the song an instant crowd-pleaser.

Beyond BPM, the AI can also detect tempo variations. In the studio, where tracks are ordinarily recorded to a click track, tempo variations are fairly unusual. But live, they are far more common. Sometimes, it's a helpful gauge of how live a recording is.

Though most of the songs in our Glastonbury data set still had small tempo variations, Kendrick’s ‘Savior’, Paul McCartney’s ‘Band on the Run’ and Kacey Musgraves’ ‘Rainbow’ all changed tempos to great effect.

What does this mean?

The key takeaways are that those song performances that were the most memorable from a musical standpoint are most different from their recorded version. That could be in tempo, mood, genre or energy. Fans don’t come to a concert to hear a record performed flawlessly. They come to see something unique. And if uniqueness is a good measure of success, Burna Boy and AJ Tracey take the prize for tailoring their music to the audience.

For music supervisors and catalogue managers, all this data can be used to build up a fuller picture of the recordings in a catalogue and help differentiate between performances of a track.

Want to talk to us about how AI can help your business? Drop us a message on Twitter, LinkedIn, email or send a message via our contact form.

Share this story