* A certain person named Prasand is in my Google contacts, along with his phone number.
* Prasand has recently sent email to me.
* Prasand is a reasonably common name in a certain English-speaking country of well over a billion people.
* A common phrase when calling someone is to say "Hi this is <name>" or "Hey this is <name>"
* Usually this comes near the beginning of the call.
* Prasand probably also has a Google account, and Google probably associates his phone number with his account, and thus can look up interesting things about him, such as his name and various words he is likely to use, by using caller ID when he calls into a Google voice number.
* Google knows my first name. Let's call me Natch.
* Speech recognition can be made more accurate if large quantities of data are available with which to build models of how language is used in context.
* Large data sets are available on the internets. Something tells me Google may even have access to large quantities of data already.
* Google even has the capability, if it wants to, to build user-specific language models.
Google, here is your challenge:
Hire an engineering director for your Google Voice team who can manage to figure out how to do the correct transcription of the following five words at the beginning of a phone call: "Hi Natch, this is Prasand."
Hint 1: you should fire whoever did the one you have right now.
> Large data sets are available on the internets. Something tells me Google may even have access to large quantities of data already.
You don't just need audio data, you also need the correct transcriptions to learn anything from it. This reduces the amount of available data significantly. And producing correct transcriptions is time consuming and expensive.
The setup:
* A certain person named Prasand is in my Google contacts, along with his phone number.
* Prasand has recently sent email to me.
* Prasand is a reasonably common name in a certain English-speaking country of well over a billion people.
* A common phrase when calling someone is to say "Hi this is <name>" or "Hey this is <name>"
* Usually this comes near the beginning of the call.
* Prasand probably also has a Google account, and Google probably associates his phone number with his account, and thus can look up interesting things about him, such as his name and various words he is likely to use, by using caller ID when he calls into a Google voice number.
* Google knows my first name. Let's call me Natch.
* Speech recognition can be made more accurate if large quantities of data are available with which to build models of how language is used in context.
* Large data sets are available on the internets. Something tells me Google may even have access to large quantities of data already.
* Google even has the capability, if it wants to, to build user-specific language models.
Google, here is your challenge:
Hire an engineering director for your Google Voice team who can manage to figure out how to do the correct transcription of the following five words at the beginning of a phone call: "Hi Natch, this is Prasand."
Hint 1: you should fire whoever did the one you have right now.
Hint 2: less AI, more common sense.