Looking for a method in order to specify usage per user using 1 subscription key per organization for speech-to-text

Question

Hi,

We are right in the middle of developing a reporting application for health care use.

Average use will be 20 concurrent users using AI Speech-to_text per customer/organization.

Technically, everything seems to be working fine. However, when using 1 key for all users, we do expect customers to ask for a detailed usage report (invoice). Creating a Microsoft account, resources and key per user seems to be overkill.

How do we go from here?

Answer 1

Hi John Keulen,
Greetings & Welcome to Microsoft Q&A forum! Thanks for posting your query!

• Implement a custom user identification mechanism within your application. Each user can be assigned a unique identifier (example: user ID) that is included in the request payload when making API calls to the Speech-to-Text service.
• Use Azure Monitor and Application Insights to track and log these custom identifiers along with the API usage metrics. This way, you can correlate the usage data with specific users.
• Create custom metrics in Azure Monitor to track the duration and number of speech-to-text sessions per user. This can help you generate detailed usage reports for each user.
• Utilize Azure API Management to manage and monitor API usage. You can set up policies to log user-specific data and track usage per user. This can also help in throttling and rate-limiting based on user-specific quotas.
• Develop a custom reporting solution that aggregates the usage data from Azure Monitor and Application Insights. This solution can generate detailed usage reports and invoices for each user based on their unique identifiers.

Hope this helps. Do let us know if you have any further queries.

---

If this answers your query, do click Accept Answer and Yes for was this answer helpful.

Answer 2

Hi,

Right now, we are using a SpeechRecognizer object based on the SDK. Purpose is real-time transcription from the default microphone (so no batch-based or file-based processing).

In my opinion, this setup does not provide any identification solutions for individual users using one common key.

Does your solution mean switching to the REST API?