Batch transcription configurations
Batch transcription configurations define how your recorded calls are transcribed after each call completes. You create a reusable configuration that specifies the transcription engine, language, result destination, and participant labels.
Legal notice and public beta
Batch Transcription Configurations use artificial intelligence or machine learning technologies. By enabling or using any of these features or functionalities within Batch Transcription Configurations, you acknowledge and agree that your use of these features or functionalities is subject to the terms of the Predictive and Generative AI/ML Features Addendum.
Batch Transcription Configurations is currently available as a Public Beta release and the information contained in this document is subject to change. Some features are not yet implemented and others may be changed before the product is declared as Generally Available. Public Beta products are not covered by the Twilio Support Terms or Twilio Service Level Agreement.
Batch Transcription Configurations is not PCI compliant or a HIPAA Eligible Service and should not be used in workflows that are subject to HIPAA or PCI.
Info
You can create and manage batch transcription configurations in the Twilio Console or using the API. This feature isn't available in the legacy Console.
Warning
Batch transcription configurations aren't available on PCI or HIPAA accounts, on accounts that use Conversation Intelligence (classic), or on accounts that have recording encryption or external storage enabled.
To create a transcription configuration using Twilio Console, follow these steps:
- In the Console, go to Products & services > Voice > Transcriptions.
- Select Create configuration.
- On the Name configuration step, enter a name and description for the configuration.
- On the Set up language & engine settings step, choose your language mode, engine, and model. See Language and engine.
- On the Destination step, configure where Twilio sends transcription results. See Destination.
- On the Participant mapping step, assign each audio channel to a participant type. See Participant mapping.
- On the Summary step, review your settings, and then select Create configuration.
The following sections describe each setting you configure in the wizard.
Language and engine settings define how your calls are transcribed. Choose a language mode, and Twilio configures the engine and speech model based on your selection. Available languages depend on the selected engine and model.
| Option | Description |
|---|---|
| Monolingual language code | Transcribe audio in a single, specific language. You specify the language code. |
| Multilingual language code | Transcribe audio that might contain multiple languages. |
| Twilio managed | Let Twilio select the optimal engine and model settings. |
Based on your language setting, the following fields are configured:
| Field | Description |
|---|---|
| Language | The language code for transcription (for example, en-US). |
| Transcription engine | The speech-to-text engine that processes the audio (for example, Deepgram). |
| Speech model | The model used by the engine (for example, Nova 3). |
Multilingual language code defaults to Deepgram with the Nova 3 model. With Twilio managed, Twilio optimizes these settings for you.
Destination settings control where Twilio sends transcription results. You must configure at least one destination method.
Send transcription results to your application using a webhook by providing a Webhook URL. The URL must be a valid endpoint that accepts POST requests. Twilio sends transcription results and related events to this URL.
You can also send transcriptions into a conversation configuration in Conversation Orchestrator for further analysis.
Use participant mapping to label each audio channel's speaker so that transcripts and analytics correctly attribute the conversation. Participant mapping applies when the associated recording configuration uses Dual channels.
Assign one channel to Customer and the other to the agent handling the call. Use Human agent for a human representative or AI agent for an automated agent.
| Participant type | Description |
|---|---|
| Customer | The external user on the call. |
| Human agent | Internal staff such as support, sales, or operators. |
| AI agent | An automated agent handling the call. |
Participant 1 corresponds to Channel 1, and Participant 2 corresponds to Channel 2. These mappings are defaults that you can override for each request when you call the Batch Transcription API.
To edit an existing transcription configuration using Twilio Console, follow these steps:
- In the Console, go to Products & services > Voice > Transcriptions.
- Select a configuration from the list.
- Select Edit on the section you want to update.
- Make your changes, and then select Save.
To delete a transcription configuration using Twilio Console, follow these steps:
- In the Console, go to Products & services > Voice > Transcriptions.
- Select a configuration from the list.
- Select Delete transcription configuration.
Warning
Deleting a transcription configuration is permanent. Recording configurations that referenced the deleted transcription configuration stop triggering batch transcription.