How Speech-to-Text billing works?

Question

How Speech-to-Text billing works?

Jefferson 0

Hi, Currently, we have 3 implementations using Microsoft AI Services – Cognitive Services. The Fast Transcript and Batch Transcript are being used as background processes. However, one endpoint and key are being used for both the Batch Transcript and the Microsoft SDK for front-end speech recognition.

We assumed that the SDK is also using the Batch Transcript for Microsoft's billing charges. We would like to confirm if this is incorrect. If it is incorrect, is there any way we can be charged less by changing how we use the SDK and be charged lower than real-time transcription?

I have read online that it's possible to use the SDK with the recognizeOnceAsync integration rather than startContinuousRecognition, as it acts like Fast Transcription and would be charged as such.

Could you please confirm this? Thanks!

SriLakshmi C 6,250 Reputation points Microsoft External Staff Moderator

2025-04-10T16:45:49.11+00:00

Hello @Jefferson,

The billing for Microsoft’s Speech-to-Text services depends on the method and API used. While Batch Transcription is intended for asynchronous processing of large audio files and is generally more cost-effective, the Microsoft Speech SDK whether using startContinuousRecognition() or recognizeOnceAsync() is billed under real-time transcription rates. Although recognizeOnceAsync() behaves similarly to Fast Transcription, it does not qualify for separate or reduced pricing and is still categorized as real-time usage.

To optimize costs, it's recommended to use Batch Transcription for non-interactive or long-duration audio scenarios. For front-end or real-time interactions, the SDK remains appropriate but may incur higher charges. Ensuring efficient use of audio input and minimizing unnecessary session durations can help reduce costs.

Please refer this for further clarification Generating real-time transcripts, Azure Speech Services Pricing.

I hope this helps. Do let me know if you have further queries.

Thank you!
SriLakshmi C 6,250 Reputation points Microsoft External Staff Moderator

2025-04-14T08:40:44.0033333+00:00

@Jefferson

Following up to see if the above suggestion was helpful. And, if you have any further query do let us know.

Thank you!
SriLakshmi C 6,250 Reputation points Microsoft External Staff Moderator

2025-04-15T09:08:31.13+00:00

@Jefferson

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thank you!
Jefferson 0 Reputation points

2025-04-16T07:06:56.5733333+00:00

Hi

Sorry for the delayed response. Your comment makes sense. I will just explore other options but thanks though.

1 answer

Your answer

SriLakshmi C 6,250 Reputation points Microsoft External Staff Moderator

2025-04-10T16:45:49.11+00:00

Hello @Jefferson,

The billing for Microsoft’s Speech-to-Text services depends on the method and API used. While Batch Transcription is intended for asynchronous processing of large audio files and is generally more cost-effective, the Microsoft Speech SDK whether using startContinuousRecognition() or recognizeOnceAsync() is billed under real-time transcription rates. Although recognizeOnceAsync() behaves similarly to Fast Transcription, it does not qualify for separate or reduced pricing and is still categorized as real-time usage.

To optimize costs, it's recommended to use Batch Transcription for non-interactive or long-duration audio scenarios. For front-end or real-time interactions, the SDK remains appropriate but may incur higher charges. Ensuring efficient use of audio input and minimizing unnecessary session durations can help reduce costs.

Please refer this for further clarification Generating real-time transcripts, Azure Speech Services Pricing.

I hope this helps. Do let me know if you have further queries.

Thank you!
SriLakshmi C 6,250 Reputation points Microsoft External Staff Moderator

2025-04-14T08:40:44.0033333+00:00

@Jefferson

Following up to see if the above suggestion was helpful. And, if you have any further query do let us know.

Thank you!
SriLakshmi C 6,250 Reputation points Microsoft External Staff Moderator

2025-04-15T09:08:31.13+00:00

@Jefferson

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.

Thank you!
Jefferson 0 Reputation points

2025-04-16T07:06:56.5733333+00:00

Hi

Sorry for the delayed response. Your comment makes sense. I will just explore other options but thanks though.

Answer 1

Hi Jefferson

It's great to hear that you are utilizing Microsoft AI Services - Cognitive Services for your implementations.

Based on the information you provided, it seems that you are concerned about the billing charges related to using the Microsoft SDK for front-end speech recognition.

When using the Speech SDK for front-end speech recognition, you are charged for both the Speech-to-text recognition request and the Language service request for Conversational Language Understanding (CLU). The billing is based on the type of service you are using and the subscription level you have.

If you are looking to reduce costs and be charged less than real-time transcription, you can consider using the recognizeOnceAsync integration instead of startContinuousRecognition. The recognizeOnceAsync method is suitable for scenarios where you need a single recognition result and does not require continuous recognition. This method is more akin to Fast Transcription and may result in lower charges compared to continuous recognition.

To confirm the billing details and explore ways to optimize costs further, I recommend referring to the official Azure Cognitive Services pricing page for the latest information and details specific to your subscription level and usage patterns.

If you have any more questions or need further assistance, feel free to ask!

Abdelaziz

Share via

How Speech-to-Text billing works?

1 answer

Your answer