We are trying to use the dictionaryfeedback API to improve the speech recognition of several business words used by our customer, but it is not working as intented. For example we are using the word finanseguro that the Speech to Text Engine is transcribing like final seguro. We have made several test changing the boostValue and it is still not working. Can someone support us to check if there is something wrong with the configuration we made.
This is the json used for the configuration.
{
"term": "finanseguro",
"dialect": "es-us",
"boostValue": 10,
"examplePhrases": [
{
"phrase": "que es un finanseguro",
"source": "Manual"
},
{
"phrase": "con cuales condiciones financio un finanseguro",
"source": "Manual"
},
{
"phrase": "aplicacion de pago a un finanseguro",
"source": "Manual"
},
{
"phrase": "cambio de tomador a un finanseguro",
"source": "Manual"
},
{
"phrase": "cambio de plazo a un finanseguro",
"source": "Manual"
}
],
"soundsLike": [
"final seguro",
"fin al seguro"
]
}
I would make sure both inbound and InQueue flows have "es-us" language (as per your original term's dialect) & I assume that is the case. Lets see if someone else have any suggestion.
what's the language of call transcript when you check under interactions view?
Also you mentioned Bot, are you checking the Dictionary Term under Bot Utterance History or Interaction Transcript ?
I dont have the transcript enable for this flow.
This is what show the utterance History. Every utterance is one of phrases used in the configuration of the DictionaryFeedback API
Good that you mentioned Bot as I don't believe dictionary terms work for bot utterances. I have seen dictionary terms in action only for call transcripts not for bot utterances. Happy to be proven wrong as keen to make use of terms with my bots