Skip to contentSkip to navigationSkip to topbar
On this page

Pre-built Language Operators


Pre-built Language Operators are Operators developed by Twilio in-house or trained using third-party AI models. This page covers the AI model used for training, Operator action, and supported languages for each of Twilio's Pre-built Operators.

The following Pre-built Language Operators are currently available:

NameAI modelOperator actionDescriptionLanguages supported
Sentiment AnalysisOpenAI GPT-4o miniClassifyClassifies call sentiment as positive, negative, neutral, or mixedAll
SummarizationOpenAI GPT-4o miniText generationCreates a summary of the conversationAll
Entity RecognitionAmazon ComprehendClassify-ExtractIdentifies and extracts entities from the conversationde-DE, en-AU, en-GB, en-US, es-MX, es-ES, es-US, fr-FR, it-IT, pt-BR, pt-PT
Agent IntroductionTwilioPhrase matchingDetects if a participant introduced themselves during the conversationen-AU, en-GB, en-US
Escalation RequestTwilioPhrase matchingDetects if a participant requested an escalation during a conversationen-AU, en-GB, en-US
Outbound Call DispositionTwilioClassifyClassifies the outcome of an outbound callen-AU, en-GB, en-US
Recording DisclosureTwilioPhrase matchingDetects if a participant disclosed the call was recordeden-AU, en-GB, en-US
Voicemail DetectionTwilioClassifyDetermines if a call went to voicemail or was answered by a humanen-AU, en-GB, en-US
Unavailable Party DetectorTwilioClassifyDetermines if the called party is unavailableen-AU, en-GB, en-US
Do Not Contact MeTwilioPhrase matchingDetects if a participant requested not to be contacteden-AU, en-GB, en-US
Non English CallTwilioClassifyDetermines if the conversation is in Englishen-AU, en-GB, en-US
Password ResetTwilioPhrase matchingDetects if a participant requested a password reseten-AU, en-GB, en-US
Call TransferTwilioClassifyClassifies if the call was transferred to another agenten-AU, en-GB, en-US

Operator Types

operator-types page anchor

Sentiment Analysis

sentiment-analysis page anchor

Action: Classify

Base Model: OpenAI GPT-4o mini

(information)

Info

Approximately, the first 90,000 words of the transcript are used to evaluate the sentiment of the conversation.

The Sentiment Analysis operator evaluates the sentiment of the conversation. The Sentiment Analysis operator doesn't analyze a specific participant of the conversation.

ClassDescription
positiveThe sentiment of the call was positive
negativeThe sentiment of the call was negative
neutralThe sentiment of the call was neutral
mixedThere was more than one sentiment during the call

Action: Text generation

Base Model: OpenAI GPT-4o mini

(information)

Info

Approximately, the first 90,000 words of the transcript are used to create a summary highlighting key points and important information from the conversation.

The Summarization operator creates a summary highlighting key points and important information from the conversation.

Action: Classify-Extract

Base Model: Amazon Comprehend

The Entity Recognition operator identifies unique entities such as locations, organizations, consumer goods, people, quantities, and times.

EntityDescription
PersonIndividuals, groups of people, nicknames, fictional characters
LocationA specific location, such as a country, city, lake, building, etc.
OrganizationLarge organizations, such as companies, government, sports teams, etc.
Consumer_goodA branded product
DateA full date (10/20/1997), day (Wednesday), month (September), or time (10:15 AM.)
QuantityA quantified amount, such as currency, percentages, numbers, or bytes

Action: Phrase matching

Base Model: Twilio

The Agent Introduction operator detects if an agent introduced themselves during a call.

The Agent Introduction operator doesn't detect a specific type of introduction. To determine whether an agent introduced themselves in a specific way, we recommend creating a Literal Spot operator that looks for the introductions that you are attempting to find.

ClassDescription
trueThe operator spotted an introduction
falseThe operator did not spot an introduction

Action: Phrase matching

Base Model: Twilio

The Escalation Request operator detects if a customer requested an escalation during a call. It captures events like a customer asking to speak to a manager or a supervisor.

ClassDescription
trueThe operator spotted an escalation request
falseThe operator did not spot an escalation request

Outbound Call Disposition

outbound-call-disposition page anchor

Action: Classify

Base Model: Twilio

The Outbound Call Disposition operator categorizes the outcome of an outbound call. It's useful for customers who are running outbound dialing campaigns and have a set of outcomes that are focused on campaign list management.

ClassDescription
GhostCallThe call was picked up and one participant does not provide any audio
IVRThe call was picked up by an automated Interactive Voice Response (IVR) system
NoContentThere was no audio during the call
OrdinaryThe call was picked up by a human and proceeded normally
StopContactingMeThe call was picked up by a human who requested to be removed from a campaign list or requested to be placed on a do-not-call list
VoicemailThe call was picked up by a voicemail system
WrongNumberThe call was picked up by a human who indicated that it was a wrong number

Action: Phrase matching

Base Model: Twilio

The Recording Disclosure operator detects if an agent informed the customer that the call was recorded. A typical use case for the Recording Disclosure operator is for outbound calls where a human agent should greet the customer before informing the customer that the call is being recorded.

The Recording Disclosure operator doesn't match a specific type of recording disclosure. If you have a specific disclosure that you want to ensure that agents use, we recommend creating a Literal Spot operator to match the required phrasing.

ClassDescription
trueThe operator spotted a participant mentioning that the call was being recorded
falseThe operator did not spot a participant mentioning that the call was being recorded

Action: Classify

Base Model: Twilio

The Voicemail Detection operator identifies if a call went to voicemail. It's useful for customers who are running outbound dialing campaigns and want to use the results to determine whether to call a specific customer back, and to assess the best time to make that call.

Unlike the Outbound Call Disposition operator, the Voicemail Detection operator only indicates whether a call was handed to a voicemail system. It does not use acoustic features to determine whether or not the call was actually picked up by the voicemail system.

ClassDescription
voicemailThe call was picked up by a voicemail system
not_voicemailThe call was not picked up by a voicemail system

Unavailable Party Detector

unavailable-party-detector page anchor

Action: Classify

Base Model: Twilio

The Unavailable Party Detector operator identifies if the called party is unavailable and the call was unanswered. It's useful for customers who are running outbound dialing campaigns and want to use the results to determine whether to call a specific customer back, and to assess the best time to make that call.

ClassDescription
NoClassBoth parties were available
UnavailableParty
  1. Customer calls a specific agent but it's not available to take the call
  2. Agent calls the customer but someone tells the agent the customer is not home
  3. Customers call the correct company but the specialized service is not available to help and the receptionist offers to set up a call back
  4. Customer needs to be transferred to someone else but no one is available to take the call

Action: Phrase matching

Base Model: Twilio

The Do Not Contact Me operator detects if a customer requested not to be contacted during a call.

ClassDescription
trueCustomer requested to stop contacting this number
falseCustomer did not request to stop contacting this number

Action: Classify

Base Model: Twilio

The Non English Call operator identifies if the conversation was in English.

ClassDescription
NonEnglishCallParticipants spoke in a non-English language
EnglishCallParticipants spoke in English

Action: Classify

Base Model: Twilio

The Password Reset operator detects if a call participant requested a password reset.

ClassDescription
NoPasswordResetPassword Reset was not requested
PasswordResetPassword Reset was requested

Action: Classify

Base Model: Twilio

The Call Transfer operator identifies if the call was transferred.

ClassDescription
NoTransferThe call was not transferred
TransferThe call was transferred

When there is audio that doesn't correspond to speech, or isn't recognized by Twilio's speech recognition engine, it will be labeled with a Non-Speech Tag. Currently, there are the following tags:

Non-Speech TagDescription
[applause]Included if a participant claps on a call
[dtmf]Included when a participant provides input via DTMF (Dual-Tone Multi-Frequency). This tag is only included when DTMF is embedded in the audio of the recording. Out-of-band DTMF is not captured
[foreign]Included when the speech recognition engine does not recognize the audio as being part of a supported language
[hes]Included when a participant says a hesitation marker like umm, uhh, or hmm
[inaudible]Included when there is unclear audio that cannot be recognized by the speech recognition engine
[laugh]Included when a participant laughs on a call
[music]Included when music is detected on a call. This marker typically shows up with hold music
[noise]Included when there are noises that are not recognized as speech
[ring]Included when there is ringing on a call. This typically shows up when a call is recorded with the record-from-ringing parameter or when a bridged leg plays ringback

To understand how to use Pre-built Language Operators on your Voice Intelligence Service, please visit the following resources: