Besides the macro drivers (usage patterns, advertiser adoption), the technology continues to improve. For example, companies like Marchex are innovating with call analytics and attribution, as explored recently.
Going back further in the value chain, there are underlying technologies like speech-to-text processing. These likewise continue to improve, boosting the capabilities of the entire downstream value chain.
With that backdrop, VoiceBase launches it’s new Insights product today at the Contact.io conference in San Francisco.* It’s also announcing that Insights will power Telmetrics‘ call tracking engine as its first partner.
Voicebase already handles recording, transcription, and analytics. Insights adds the additional dimension of predictive intelligence. Its automated approach can also lower costs, co-founder Jay Blazensky told me.
VoiceBase does this by using machine learning to score calls. It develops models based on historical calls that are used as a quality benchmark. They’re analyzed for up to 10,000 variables like keywords and voice tone.
“We start with those tagged calls that highlight a signal or event that classifies the call,” said Blazensky. “We upload hot leads and cold leads, and our machine learning will figure out how to recognize them.”
The point is that accurately detecting call quality is the name of the game in call analytics, as we’ve learned from Marchex, Invoca and others. Leads are scored for marketing, as well as call routing and call center efficiency.
As an API — agnostic to voice markets — Blazensky names call centers, voicemail applications and PBX companies as users. But it’s not just about calls… the addressable market is anywhere human voice resides.
Acadamia and presentation transcriptions are another killer app, Blazensky says (or a BIA/Kelsey conference). VoiceBase is working with C-Span for session recordings, and with Nasdaq for corporate earnings calls.
But the value isn’t just on-demand recordings, it’s having an accurate transcription by which to build a keyword index. That brings a long-missing capability to audio: searchability. Think: Podcasts and Presidential addresses.
“It becomes a repository of information that is now queriable,” said Blazensky, who refers to this collision of data and voice as “Big Voice.”
Similarly, I’ve theorized for years that Google’s work in voice (i.e. voice search, Google Voice), is a trojan horse for refining its speech to text processing, so that it can make video as effectively indexable as text is.
That’s the first step towards making video more searchable and thus monetizable through its paid search model. It would open up all kinds of ad inventory, most of which is already under it’s own roof via YouTube.
Back to calls, As we’ve said recently, even in a digital age (and the rise of messaging apps) the biological and sociological drive to talk with another humans isn’t going away. This is especially true in commerce.
And it shines most in high-value product categories where there is inherent complexity. Think: auto, healthcare, and financial services. The dollar values of these products makes their voice-based lead flow so valuable.
As for VoiceBase and the Insights platform, it will get its chance to prove whether or not it lives up to the promise of automating the nuances of call scoring. We’ll be watching closely.
*If you’re at Contact.io, I’ll be speaking on a panel at 2:00 today: Why Now? The Revolution Behind Contact: Understanding What Consumers Are “Talking” About.