Standard Speech to Text and Text to Speech Actions in Flow
Summary
Salesforce introduced two new standard Flow actions, Speech to Text and Text to Speech, powered by Agentforce, that enable native voice capabilities without custom code. These actions allow Salesforce teams to transcribe audio files stored in Salesforce or generate natural spoken audio from text directly within Flows. They support multiple languages, various audio formats, and provide flexible options like choosing transcription models and voice parameters. However, these actions require specific Agentforce entitlements, consume Flex Credits, and have usage limits, making them suitable for scalable automation with cost considerations. Salesforce professionals can leverage these new building blocks to enhance user interactions and automate voice-related processes seamlessly.
Takeaways
- Use Speech to Text action to transcribe audio files stored as Content Documents in Salesforce Flows.
- Leverage Text to Speech action to convert text inputs into spoken audio, optionally saving as Salesforce Files.
- Monitor Flex Credit consumption as both actions use separate metered credits with usage limits.
- Ensure Agentforce entitlement and edition compatibility before deploying these voice actions.
- Build additional logic to store transcriptions since Speech to Text only returns transcripts without saving.
Salesforce quietly rolled out two new standard invocable actions that bring voice capabilities straight into Flow: Speech to Text and Text to Speech. Until now, converting audio to text (or the reverse) inside Salesforce meant either building a custom LWC around the browser's Web Speech API or wiring up an external service integration yourself. These new actions skip all of that. They are native, composable building blocks powered by Agentforce's speech infrastructure, and they are ready to drop into any Flow without writing a line of code. In this post, you will learn what each action does, what inputs it expects, and a few things worth knowing before you build them into your automation. Where These Actions Are Available These actions aren't just Flow actions. They are also available as Agentforce agent actions, and they aren't available on every org by default. They are part of Agentforce's Speech Foundations, which requires an Agentforce entitlement.