Why am I getting a quota error?

Rich Hawksworth 0

I'm using Azure TTS and getting the following quota error:

"You have reached the quota with your free-tier (F0) Speech resource. To continue to create audios with neural voices, switch to a standard paid resource, or upgrade your free-tier resource to a standard (S0) tier. "

I'm using the Free (F0) tier which has a limit of 500,000 characters/month. I've used maybe 30,000 billable characters before getting this error, so I'm nowhere remotely close to the max. Individual file sizes are also very small. < 200 characters.

I have to be missing something, right? I'm not adverse to pay-as-you-go pricing, but my workload should be easily manageable within the F0 package.

Thanks in advance for any suggestions.

VasaviLankipalle-MSFT 14,911 Reputation points

2024-03-06T02:28:51.6733333+00:00

Hello @Rich Hawksworth , Thanks for using Microsoft Q&A Platform.

Is it possible to share the documentation you are following that has mentioned this " Free (F0) tier limit of 500,000 characters/month".
Rich Hawksworth 0 Reputation points

2024-03-06T11:59:37.0233333+00:00

This page provides an overview of Speech AI pricing: https://azure.microsoft.com/en-us/pricing/details/cognitive-services/speech-services/

This same limit is indicated within my account dashboard where there is a side-by-side comparison of the f0 and S0 tiers. The Feature "500K char/month - Neural Text-to-speech" is ticked for the the f0 tier.
VasaviLankipalle-MSFT 14,911 Reputation points

2024-03-08T00:40:34.45+00:00

Hello @Rich Hawksworth , thanks for sharing this information.

In the Azure portal, try to Analyze Metrics to find the sum of characters. At the left side of your resource, you can choose Metrics and set time period to last 30 days to check the sum chars.

Please try this and let us the know the Synthesized Characters (Sum) count.
Rich Hawksworth 0 Reputation points

2024-03-08T01:17:56.2233333+00:00

Thanks for this. The metrics indicate I've used 506k in the last 30 days. I understand what a billable character is -- how it includes text characters, punctuation, white spaces, and some parts of the ssml. I guess I must not understand correctly how these billable characters are triggered.

For example, let's say I am working on a file that has 300 billable characters. Am I getting "billed" every time I play that file in the speech studio editing window? As an audio editor, I will listen to a passage numerous times, working on intonation, pacing, etc. before I'm satisfied that it's ready for export. I would think that the 300 billable characters would be absolute -- meaning I would get billed for 300 characters regardless of how many time I listen to the file. But am I really getting billed 300+300+300+300, etc. to account for every time I play it?

Of course, this is a matter of pennies when it comes to 300 characters, but my employer needs to understand the implications of working on 3M "billable" characters or 30M "billable" characters. For context, those 506K billable characters reflected on my account was somehow accrued from fine-tuning only about 30k total characters (including text, white space, ssml, etc.). So, somehow that 30k became 500k and I'm struggling to understand how. I would greatly appreciate it if you could clarify how billable characters are triggered.
VasaviLankipalle-MSFT 14,911 Reputation points

2024-03-08T01:43:13.1966667+00:00

Hello @Rich Hawksworth ,

For Text to Speech: usage is billed per character. Please check the definition of character in the pricing note here.
Rich Hawksworth 0 Reputation points

2024-03-08T02:57:58.02+00:00

Precisely what does “converted to speech” mean? If I have 100 billable characters in a file and I press play and listen to the string, then press play and listen to it a second time, have I just used 200 billable characters?
VasaviLankipalle-MSFT 14,911 Reputation points

2024-03-08T23:39:15.8066667+00:00

Hello @Rich Hawksworth , when a text file is "converted to speech", it means that the text is synthesized into an audio file that can be played back as speech.

Regarding your second question, TTS services typically charge based on the number of characters in the input text that is being synthesized into speech.

I have reproduced the same and sharing my observations, once the speech is generated, the "Save" button is grayed out.

If you simply replay the audio without making any changes to the text, you will not be charged again. However, if you edit the text or use tune options in the right panel and then save those changes, you will be billed for the additional characters.

Please note that phonemes and pitch, are counted as billable characters.

I hope this helps.
Rich Hawksworth 0 Reputation points

2024-03-09T01:04:57+00:00

Thank you. I've been experimenting a bit and it seems like I also get billed (for all the billable characters in a file) when an audio file is exported and again when it's downloaded. This snapshot here reflects a (previously downloaded) file that I reexported and redownloaded ... no changes to the file itself.

You've been very helpful. Thank you.
binarystar 0 Reputation points

2024-04-30T17:58:34.5066667+00:00

i met the same issue
binarystar 0 Reputation points

2024-04-30T18:03:54.3133333+00:00
hello， i met the same issue，i just registered a new account，but when i tried to use the TTS service，it shows this “You have reached the quota with your free-tier (F0) Speech resource. To continue to create audios with neural voices, switch to a standard paid resource, or upgrade your free-tier resource to a standard (S0) tier. Connection ID: 77d0c712-071a-11ef-b380-8dcc8f81f370”

and i can't found the using history and the quota left.

Share via

Why am I getting a quota error?