Radio - The Next Frontier For Artificial Intelligence - Radio Ink

• Updated on: 2017-04-21
• Read original article here

(By Paul Cramer) The term artificial intelligence often evokes images of sentient computers from Star Trek or cyborgs from The Terminator. While those are extreme examples of artificial intelligence or AI, the Oxford Dictionary gives a more pragmatic definition of AI as computers that have the capability to perform tasks that normally require human intelligence, such as visual perception, speech recognition, and decision-making.

AI is already beginning to transform how people work and live, from Tesla’s Autopilot-enabled self-driving cars; to the Alexa, Cortana, Siri, and Google assistants; to smartphones that scan email inboxes and provide automated alerts on everything from flight delays to package deliveries. The technology is enabling better and more relevant user experiences that would be too time-consuming and inefficient for human labor to deliver.

These same AI-derived benefits are now coming to broadcast radio, and will similarly deliver newfound value for both the industry’s programming and sales departments by capturing and enhancing live broadcasts.

Do it live As one of the first forms of mass media, radio has both benefited and suffered from its status as a live medium. The earliest news broadcasts were all live. Some of the first music on the radio consisted of broadcasts of live musical performances, such as the broadcast of the Grand Ole Opry, which began more than 90 years ago at WSM Nashville.

Indeed, the beauty of radio is that it’s live, and as such, programming can changed in near-real time based on call-ins, breaking news, and unfolding events. The downside to being live is that radio is the ultimate perishable product. The minute a broadcast leaves the transmitter, it vaporizes into the surrounding airwaves if it’s not consumed in real time.

Opening the on-demand doors Radio’s real-time perishability began to change for the first time in 2004 following the marriage of the Apple iPod to a RSS feed, resulting in the coining of the term podcasting. For the first time, radio didn’t need to be delivered and consumed in a linear fashion. Instead, the listener could time-shift and listen to a show, a play-by-play broadcast, or an interview whenever it was convenient, or on-demand.

Fast forward to today, and there are more than several-hundred thousand podcasts on iTunes alone. This presents a challenge for curation and content discovery. Moreover, with the growing popularity of personal assistants like Alexa and Google Home, many users no longer are utilizing keyboards to search through content offerings. Exposing radio stations’ on-demand, evergreen, and podcast content is currently relegated to “tag-based” searching, where discovery is dependent on a topical tag, or keyword, that a producer appends to the audio file.

The tag produces a search result, and then the listener must sample the content to see if it’s what he or she was actually looking for. This is the modern-day audio equivalent of using a microfiche machine to discover new content. It’s time-consuming and frustrating because it often does not produce the same confidence in search results that contextual search does. This is where AI will begin to play a major role in exposing stations’ on-demand content to new and repeat audiences.

AI future-proofs radio AI engines can now ingest any live or previously recorded broadcast content, and by leveraging  natural language processing, they can provide a full transcription of what was said in the broadcast. This opens up a world of new possibilities, enabling for the first time in history the capability to contextually search an entire archived broadcast.

Now, the entire content of an interview, morning-show bit, call-in program, or play-by-play sports broadcast can be exposed to search-engine algorithms. By generating a full transcription, a program and its content can be indexed and subjected to organic Web or voice searches and then presented to the listener based on relevancy scores.

The beauty of AI is that a station’s evergreen content can be discovered more easily, and when coupled with a digital-audio ad-server, it can be monetized repeatedly into the future. Therefore, a listener who searches for a Tom Brady interview today—or two-months from now—can automatically be cued to the segment of the broadcast where Tom Brady is on the air. This also future-proofs on-demand audio by allowing it to be discoverable by voice search in the growing number of connected-home and connected-car applications.

AI is also providing immediate benefits for the radio industry’s sales and revenue. When a live radio program is ingested and processed in the cloud using AI, the resulting transcription can provide immediate and actionable insights for sales teams, media buyers, and advertisers. Such a searchable transcription allows every piece of advertising—whether spot, mention, endorsement, or live-read—to be tracked and verified in near-real-time. This creates a whole new level of transparency that broadcasters can use to get credit for embedded media that was previously difficult, if not impossible, to track. Now, instead of receiving a log or affidavit-based campaign recap, advertisers can receive a proof-of-performance report for the entire campaign, including spot and live-mention reports. Moreover, this report can include automated air-checks of each mention and spot, impression delivery summaries, detailed audience reports, and much more. It can even present and give credit for earned media, such as when an advertiser is mentioned within jock chatter, a promo or a sweeper. This enables a better way to service advertising clients and deliver dynamic dialogue that’s more akin to online media than to offline media.

Generating near real-time performance reporting and automated air-checks is just one way AI will transform radio sales. This technology will also provide improved workflow efficiencies to the industry, just as AI has made tasks more efficient in other aspects of our lives. Searching through volumes of programming on the station skimmer will become akin to using a microfiche machine from years gone by. AI will make finding station copy, a voiced read or a spot that ran out of flight as easy as searching on Google. It will provide new ways to prospect for clients and to track competitive campaigns. It will even be able to put radio on even footing with digital media, where real-time campaign metrics can be used for attribution tracking and modeling.

As AI begins to truly be embraced by the broadcast radio space, the creation of new advertising insights will be just the beginning. Just as occurred with the Web, once content is digitized and indexed, the possibilities are endless. Through radio’s integration of AI, broadcasts will no longer vaporize, but instead will become “dimensionalized,” i.e. made available for a host of new distribution and revenue opportunities. Paul Cramer is Managing Director of Enterprise Radio Solutions for Veritone Media and can be reached at[email protected].