According to research firm Strategy Analytics, 86.2m smart speakers were shipped in 2018, up from 32m the year before. Meanwhile, rival Canalys reckons that the global install-base will be 207.8m by the end of 2019. Oh, and the same company thinks that China has just overtaken the US as the biggest market for these devices.

So much for market stats, but what’s coming next in terms of the features for smart speakers, and what that means for how we interact with music and musicians through them? A panel at the Midem conference today offered some predictions.

One of them was the potential for purposefully-inaccurate metadata: artists or labels deliberately mis-tagging their music in the hope that it’ll have a better chance of being picked up by the algorithms of voice assistants like Alexa, Google Assistant and Siri in response to popular queries.

“What we’ve seen with UGC… is that people either try to add data that spoofs, or that is a different genre,” said Scott Ryan, VP of Nielsen Music / Gracenote Music. “To basically find ways to intentionally try to goose the playback, but unfortunately missing the accuracy, and the people that ARE looking for their music.”

(Nobody’s calling this ‘metabait’ yet, but Music Ally is more than happy to coin the term. Pass it on!)

Benoit Rébus, head of global innovative partnerships at Qobuz, said that while streaming services are capable of fleshing out patchy metadata – “at Qobuz we have a team that enriches metadata” – there’s also an important role for labels and artists to provide as much metadata as possible. Accurate metadata, not metabait.

One topic that arose during the panel: the fact that in the west, the smart speakers market (and the voice assistants that run on them) belongs to Amazon, Google and Apple, which poses obvious competitive concerns for independent streaming services like Spotify.

Can Spotify create a better voice-assistant for music than Alexa, Google Assistant and Siri? Can it afford not to?

“Can Spotify build something that’s better? Absolutely. Will they? Who knows?” said Darryl Ballantyne, CEO of LyricFind, adding that Spotify must be concerned by the fact that the tech giants may be sitting in between its service and the raw voice-queries posed to smart speakers.

“If you’re just getting it translated by Google or Amazon, is that a threat to your business? Is that something you need to be on control of? Can they afford NOT to be in control of it?” he said.

Ryan also raised the prospect that if people are getting better experiences with the native streaming service for a smart speaker – Amazon Music Unlimited for Echo devices, or Apple Music for HomePods for example – than they are with rival services, there are clear anticompetition issues. Something Spotify has made part of its current EC complaint against Apple.

“Voice control like that? We’re starting to hear a lot about antitrust. I think that’s going to be one of the key areas,” said Ryan. However, Rébus claimed that Apple is preparing to open up its Siri API to other music services, which would be important for the likes of Spotify (and Qobuz).

The panel also talked about the potential for even smarter, more personalised music services delivered through smart speakers. Cohen cited a new (as-yet unannounced) service his company is working with in the US, which will look at the weather and time of day, then create suitable playlists as an example of how context could play an even greater role.

Ballantyne was enthusiastic about where voice-control goes next. “The cheesy stupid answer is ‘everywhere’. The reason for that is you essentially have unlimited screen real-estate through voice search,” he said, citing the large range of possible voice commands.

“You could never fit that in to a regular screen interface for a music service. But when it’s just voice technology using databases and knowing which dataset to go to to answer those queries, you have limitless shelf space for search and discovery.”

But metadata remains the key challenge, and one that could throw up more headaches as time goes on. “It’s the need for clean data. The need for context. What’s popular and what’s new?” said Ryan.

He added that the sheer volume of new music, as well as the amount of music that’s crossing borders, are trends that bring their own challenges.

“It’s only created that many more opportunities for confusion. There’s something like 40 bands called Exile, so how do I know which is the right one?”