The context we set with James was that when we say ‘voice’ we think ‘voice activation’ but this is also a huge year for social audio, kickstarted by the quick growth of Clubhouse and followed by the likes of Facebook, Twitter, Spotify and more with social audio feature launches. We threw this into the melting pot with James, who raised some super interesting areas of challenges and opportunity for start-ups, from realtime speech processing for recommendations and moderation to sentiment analysis as well as helping people learn and understand how to interact with voice assistants.

Social audio offers a way for artists to connect with fans in a way that's more authentic than other social media platforms. The scarcity factor - no record of the experience once it's ended - makes social audio feel unique and desirable to fans. It's also particularly interesting to artists who don't want visual media to be their main way of connecting with fans.

But that does come with challenges. The infrastructure isn't completely developed yet. For example in Clubhouse there is no DRM protection (although recent launches by big players will put the wheels in motion). There are also discoverability issues: how do you get your fans in that social audio platform? How can you stand out above the noise in-platform, and keep them listening to you?

This bridges to more technical issues in discoverability on audio platforms in general. Firstly, social audio is competing with the advanced ecosystem of music playlists and clever filtering systems. Secondly, social and podcast audio platforms have ecosystem issues, in that the metadata frameworks are not nearly advanced enough to enable rich recommendation and highly personalised discovery.

However, there is progress in this area, especially with speech to text technology which can then enable word and sentiment searching and recommendations across podcasts. Speech to text and text to speech is a hugely important area of development. For example, recommendations of social audio spaces mean monitoring a space, converting, processing, and recommending – all in real-time. Features like live keyword detection are needed for moderation as well as recommendations. Advanced ML systems will allow platforms to automatically spot rooms that are scaling, and recommend them to others in personalised ways.

Sentiment analysis is another key area to be exploited. Having data on hand on whether a room or space is ‘happy’ or ‘sad’; whether a debate is happening or not; if the conversation is heated or lacklustre – anything which helps panellists, moderators, and listeners know what the energy of that room is will help improve spaces, and share them with others.

The final problem I want to touch on - which of course is also an opportunity - is the ‘training’ issue in the voice activation world. You need to help people understand how to interact with their smart speakers and remind them that this is a great way of searching for music and social audio. Catering to this is the rising trend in social and voice activation: telling people in marketing channels what they can do with voice activation and how.

The evolution of no code solutions. Platforms like Voiceflow allow users to prototype concepts, launch, and test. No code solutions democratise the ability to build voice activation apps and features in social audio platforms.

The third and fourth layers – management and analytics. Creators will benefit from tools to more intelligently host spaces, and data to analyse on how to keep audiences engaged and happy. I'd love to see a Hootsuite for Twitter Spaces or Clubhouse!

I also think we'll see brands and businesses start to take social audio more seriously, as Vixen Labs has been seeing over in the main voice-first technology space for a while now with strategy, search, commerce, and apps. As social audio is very much about person-to-person connections, brands are going to have to think very carefully about who represents them, what messages they want to share with the world, and who they want to connect with. There is still some first-mover advantage left to claim in much of the social audio technology space.