Skip to main content
Prerequisite: Please verify amount of channels in the audio file before proceeding.

​
Once obtained your API key, you can start using the API by following the Quickstart Guide.

To get the most out of our API, please refer to the Understanding Insights & Metrics section.

​
Basic Parameters and Configuration

language_code: Choose the language of your call recordings (for example, en-US).transcript_model: Select which speech-to-text model to use. Our default ‘descriptor’ model is optimized for call centers. We don’t send any audio and voice data to third-party services.emotions_model: Enable this to analyze vocalized customer emotions in calls. Set to null if you don’t need speech emotion recognition.emotions_alignment: Get more detailed emotion analysis by matching specific phrases to emotions. Set to null if not needed.emotions_diarization: Track who said what and their emotional state throughout the call. Works best with emotions_alignment. Set to null if not needed.channels: Tell us how many separate audio streams are in your recording (for example, separate channels for agent and customer). We can handle up to 8.insights: Choose what information you want extracted from calls - like conversation summaries, main topics discussed, or call categorization. You can customize the AI service used for each type of analysis.sentiment_llm_provider: Choose which AI service analyzes the emotional tone of conversations. We use Azure by default, and we also support on-premises LLMs.metrics: Select which measurements you want: ‘speech’ for conversation flow metrics, ‘emo’ for emotional analysis metrics. Skip ‘emo’ if you don’t need emotion tracking.webhook: Provide a web address where you want us to send the results when analysis is complete.

​
Response Overview

Once processed, the response will contain the following information:
  • Call Transcript: Text transcription with timestamps for each speaker.
  • Emotion Analysis (if enabled): Emotion markers per phrase, including sentiment and intensity.
  • Insights: Extracted insights such as summary, topics, customer issues, labeling, CSAT score, agent actions, and customer requests/questions.
  • Metrics: Calculated metrics including speech and emotional metrics as configured. For more details, please visit api-reference. This JSON response can be integrated into your system or visualized in a GUI for further analysis at https://demo.descriptor.ai/result/{result_id}.
If a webhook URL was provided in the configuration, the system will automatically send the completed results to the specified endpoint.
For your convenience, we’ve prepared a demo app that showcases the capabilities of Descriptor.AI. It’s not required to run the app to use the API, but it might help you to understand how to use the API and what results you can get.Please refer to the section above for more details regarding the configuration and parameters.
You need to have an API key to use the demo app.
We don’t provide the app for production use, so please don’t use it in a production environment.
Demo is available at demo.descriptor.ai.