Voicera Announces Progressive Attention Feature to Balance Transcription Speed with Accuracy and Highlighting
Voicera CEO Omar Tawakol started his interview earlier this month saying the company organizers its offering around two things, “Stop taking notes. Let Eva do it for you.”; “You can’t attend every meeting, but Eva can.” The first point relates to your focus during meetings. If you are furiously scribbling or typing notes, you are probably missing important comments and at least interrupting the natural conversational flow. Second, there is the fear of missing out (#FOMO) on content in meetings that you cannot attend for one reason or the other. Eva is Voicera’s assistant which offers a meeting transcription and note taking service for business users.
Many people are aware that automated speech recognition (ASR) and transcription has improved tremendously over the past few years. However, some things may not be obvious. First, transcribing a meeting with multiple speakers is much more challenging than simply capturing what one person said to your voicemail. Second, there is a tradeoff between speed and accuracy when executing transcriptions. That tradeoff is more acute when you are looking to highlight notes or otherwise annotate what was said.
Balancing Speed, Accuracy and Intent Extraction
Voicera’s new Progressive Attention AI approach offers a new method to balance speed, accuracy and intent extraction from an extended conversation. A announcement describing the progressive attention feature offers this summary:
“The human brain has always had the luxury of constantly scanning its total environment while not wasting precious brain resources on every noise. Then when the brain senses something important, it allocates more attention to carefully understand a particular input from the environment. Without it, people wouldn’t be able to both engage in high level conversations and thoughts while also staying safe from threats. Most AI systems suffer from this same ‘sophie’s choice’ of expensive time-consuming processing OR lower accuracy. Progressive Attention AI solves this, by allowing a fast analysis of the environment to identify relevant inputs followed by very deep and focused processing of important moments that can yield a much higher accuracy.”
Tawakol said that Voicera arrived at this approach after reviewing data from over “10,000 real-world meetings” and completing discussions with 300 users over two months. The announcement elaborates that:
“From both methods we established that users did not want to relive an entire meeting. They do not want to review a full transcript because that is not an efficient use of time…Transcripts are a safety net, but highlights…are the notes that can be used to share insights, guide follow up and ensure that the customer is heard properly. Our focus quickly shifted from accuracy in full transcripts to the accuracy and readability if highlights.”
Ensemble Goes Deeper to Extract Highlights
The engine behind Progressive Attention AI is called Ensemble according to Tawakol. Voicera still captures transcripts in real time and having tried it recently, I can say it is pretty accurate. Tawakol said that there are typically two engines capturing the real time transcription and then Ensemble provides deeper analysis to extract the meeting notes highlights. The highlights take longer to process and are typically emailed or available in the mobile app about 30 minutes after a meeting concludes.
What is immediately obvious is that the highlights are categorized into notes, decisions, next steps and action items. If Ensemble categorized the note incorrectly, you can update it in the web console. A nice feature is that the note is linked directly to a specific point in the recording so you can click to listen to just that portion of the meeting that requires more attention. You can then share that information with meeting participants and others to offer a quick summary of the meeting which streamlines post-meeting workflow.
Making Meetings a Little Less Cumbersome
Voicera may make meetings more effective, but in my brief experience it really makes them less cumbersome. You don’t have to take detailed notes and can focus on the conversation more. You have the transcript if you need to pull out a detail. The notes help you quickly zero in on key points of the discussion that may require further attention.
Previous tools of this type focused primarily on customer service and indeed some Voicera customers are using the service for that function. But, Tawakol says that about 60% of users today occupy internal facing roles and do not interact with customers. This is a market that was simply unaddressed in the past.
Voicera offers four levels of service. There is a premium service that employs Ensemble progressive attention in all meetings an unlimited transcripts for $32 per month. These transcripts all include error correction. The $16 per month pro tier doesn’t include error corrected transcripts but uses Ensemble for highlights for unlimited meetings. There are also $8 per month and free options that have limitations on highlights and transcript volume. You can find out more here.
Follow @bretkinsella Follow @voicebotai
Voicera Lands $20 Million in Funding, Acquires Wrappup, Launches Mobile Apps