如何在Dialogflow V2beta1中区分文本/音频来源及不同终端机器人

Got it, I've run into this exact issue before when migrating from Dialogflow V1 to V2beta1. The good news is there's a reliable way to distinguish between audio-based (Alexa/Google Home) and text-based bots, even though the old numeric message enum from V1 is gone.

How to Distinguish Bot Types in Dialogflow V2beta1 #

• Check the originalDetectIntentRequest.source field
  This string field directly identifies the originating platform. For example:

  • "google" for Google Assistant devices (including audio-only ones like Google Home)
  • "alexa" for Alexa-enabled voice devices
  • "dialogflowMessenger", "slack", "telegram" (text mode) for text-based bots that support Rich Responses

• Dig into originalDetectIntentRequest.payload for granular device details
  For audio platforms, the payload adds context to confirm device capabilities:

  • For Google Assistant, look for payload.surface.capabilities — if it includes actions.capabilities.AUDIO_OUTPUT, you can be sure it's an audio-only device that can't render Rich Responses.
  • Alexa's payload follows the Alexa Skills Kit request format, with fields like request.type (e.g., "LaunchRequest") that clearly signal voice-first interactions.

• Build response logic with source mapping
  Add a simple check in your webhook code to categorize platforms and serve appropriate responses:

  const isAudioOnlyBot = (source) => {
    return ["google", "alexa"].includes(source);
  };
  
  const supportsRichResponses = (source) => {
    return ["dialogflowMessenger", "slack", "telegram"].includes(source);
  };
  

  Use these functions to skip Rich Responses for audio bots, and include them for text-based platforms.

Key Note #

V2beta1 replaces V1's numeric message enum with platform-specific string sources and payloads. It's more verbose, but it gives you finer control over different device types beyond just audio vs text.

内容的提问来源于stack exchange，提问作者Jefftopia