Use Case
Citi Bank
1. Surveillance group (speech analytics for making sure their traders aren't being sketch)
2. Consumer arm (B2C contact center audio)
Send me an email
I will send you something right away. What exactly do you want to be included in that email?
It'd be much easier to show you on a quick zoom call.
Name the 2 biggest Open Source ASR options
Kaldi
Whisper
What languages does Nova support?
English and Spanish
CallTrackingMetrics
CTM is a call-tracking company providing marketing attribution and conversation intelligence for calls. They initially signed on with Deepgram in 2020 for batch traffic only but are now looking to make a big splash in the real-time market and move most of their real-time traffic from Google over to Deepgram.
We're happy with Google
- Google customers make our best customers
- 1/3 of the cost
- Our accuracy outperforms theirs
- Much faster
- We have dedicated reps for our customers to help with any issues
What makes Rev different than other ASR vendors?
They also have a human labeling business that feeds their ai models
Why would a company want an On-Premise deployment?
- Data security
- No network latency
- Manage their own machines
Podsights
Podsights transcribes podcasts to analyze the advertisements in them, to ensure that their clients are being represented well in the ads and that the copy is being read correctly.
Not interested
Is speech recognition something you’re not using currently? (get more context)
So you don’t process any audio data? I see you’re doing X on your website, is that not true? (Call out your research)
Where do we lose to Nuance?
- Medical
- Contact center use cases that want full solution
Walk me through the whole model training process
Will to judge
Cquence
They provide meeting summaries using their ML model it to fill out follow up templates. Their target customer is B2B sales reps like us.
Why are you better than Google or AWS?
- Architecture (DNN)
- Accuracy
- Speed & Cost come from model efficiency
- Dedicated support
- Continuous improvement
Which vendors can go On-Prem
Google, Nuance, Whisper, Assembly, Rev, Kaldi, Voci
Describe LLM token pricing for our new understanding features
- Will to judge
OfOne
We've built our own ASR
Are you able to do real-time?
Cost to maintain, host, the opportunity cost of engineers working on it.
Name the best way we can get a meeting with users of:
AWS
Whisper
Assembly
Rev
Google - General Accuracy / Price
AWS - On-prem, price, accuracy
Whisper - Real-time, accuracy maybe, price maybe
Assembly - price Accuracy latency
Rev - accuracy price latency
UniMRCP is an open source cross-platform implementation of the Media Resource Control Protocol (MRCP), which is a protocol for PBX's to communicate with ASR and TTS engines.
It's used in IVR use cases.