Hey there! This channel allows you to ask Kapa.ai for assistance. But it looks like you haven't connected your Discord profile to your Deepgram account yet. You can do so at https://community.deepgram.com - by doing so, you'll be able to ask questions in #1115960287183511643 if you need human assistance!
#400, message='Invalid response status
midnight turretBOT
untold bayBOT
Hi @wide dragon. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
Thank you for your question about Deepgram's nova-3 model and the 400 error you're encountering. I'll do my best to address your concerns based on the information provided in the knowledge sources.
The 400 Bad Request error you're experiencing could be due to a few reasons:
-
Model and Language Combination: Make sure that the nova-3 model supports the language you're specifying. According to the Reference > Errors page, one possible cause for a 400 error is:
{ "err_code": "Bad Request", "err_msg": "Bad Request: No such model/language/tier combination found.", "request_id": "[unique_request_id]" }Ensure that the language you're using is supported by the nova-3 model.
-
Query Parameter Format: In some cases, using incorrect parameter formats can cause 400 errors. For example, in a discussion about Streaming WebSockets : Invalid query string (400 Bad Request), it was noted that boolean parameters should use "true" or "false" instead of "1" or "0". Your query string looks correct in this regard, but it's worth double-checking.
-
Rate Limiting: Although your error message doesn't specifically mention rate limiting, it's possible that you're hitting a concurrent request limit. According to a comment by a Deepgram maintainer, different models have different rate limits:
- For Nova, Base, and Enhanced, the rate limit is 100 concurrent requests.
- For Whisper, the rate limit is 15 concurrent requests with a paid plan and 5 concurrent requests with the pay-as-you-go plan.
While nova-3 isn't specifically mentioned, it's likely subject to similar limits.
- Intermittent Issues: Since you mentioned that it works most of the time but fails during important tests, there could be intermittent issues with the service or your connection.
To troubleshoot this issue further, I recommend the following steps:
-
Check your Deepgram logs for the specific request IDs when the errors occur. This will help Deepgram support identify the exact cause of the problem.
-
Implement an exponential backoff retry strategy for your requests. This can help mitigate transient errors, as suggested in a comment by a Deepgram maintainer.
-
Verify that your account has the necessary permissions and access to the nova-3 model.
-
If the issue persists, contact Deepgram support with detailed information, including request IDs and the specific times when the errors occurred.
If you need further assistance or have more specific questions, please don't hesitate to ask. It would be helpful if you could provide any error messages or request IDs associated with the 400 errors you're encountering.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
wide dragon
@untold bay How can i contact deepgram? I think support is only available for premium plans.. could you provide me support email?
untold bayBOT
Let me check the knowledge base for you again. One moment! :books:
Based on the information provided in the knowledge sources, I can offer you several ways to contact Deepgram, depending on your account type and the nature of your inquiry:
-
Community Support:
If you're using a free, growth, or pay-as-you-go plan, you can get community assistance through: -
Premium or VIP Support:
If you have a Premium or VIP Support Plan, you can find contact details on your Console dashboard. -
General Inquiries:
For general inquiries or if you're unsure where to start, you can use the Contact Us page on the Deepgram website. -
Security-related Inquiries:
For security-related questions, you can email [email protected]. -
Developer Relations:
In some cases, the Developer Relations team may be able to assist. Their email is [email protected], as mentioned in a GitHub discussion response.
While there isn't a general support email provided in the knowledge sources, the best course of action would be to start with the community support channels or use the Contact Us page. If you have a specific issue related to your account or API usage, providing details through these channels can help route your inquiry to the appropriate team.
Remember, as mentioned in the Deepgram pricing FAQ, detailed troubleshooting and on-demand support typically require an ongoing support contract with Deepgram, which is available with Enterprise plans.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
wide dragon
@untold bay 'dg-error': "Invalid 'tier' value of 'nova-3'." this is the error mentioned in the headers of the request
untold bayBOT
Let me check the knowledge base for you again. One moment! :books:
Thank you for providing the specific error message you're encountering. This error is related to the use of the 'tier' parameter, which is actually deprecated in the current Deepgram API. Let me explain the issue and provide a solution based on the information from our knowledge sources.
The error you're seeing, "Invalid 'tier' value of 'nova-3'," indicates that you're using an outdated method to specify the model in your API request. According to the Tier (Deprecated) documentation:
Deprecation warning. Please use the
modelsyntax outlined in the Models & Languages Overview.
This means that instead of using the 'tier' parameter, you should be using the 'model' parameter directly. Here's how you can correct this:
Instead of using:
tier=nova-3
You should use:
model=nova-3
For example, your API request URL should look like this:
<https://api.deepgram.com/v1/listen?model=nova-3&language=en&smart_format=true...>
This change aligns with the current Deepgram API structure as described in the Models & Languages Overview documentation. The nova-3 model is specifically mentioned there:
<https://api.deepgram.com/v1/listen?model=nova-3>
By making this adjustment, you should be able to resolve the 'dg-error' you're experiencing. Remember to update any other instances in your code where you might be using the 'tier' parameter for specifying models.
If you continue to experience issues after making this change, please don't hesitate to ask for further assistance.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
wide dragon
@untold bay url='wss://api.deepgram.com/v1/listen?model=nova-3&language=en&smart_format=true&filler_words=false&encoding=linear16&sample_rate=16000&channels=1&endpointing=true&interim_results=true&utterance_end_ms=1000&vad_events=true', ofcourse i'm not using any tier, im using model only.
maybe it is due to rate limits?
untold bayBOT
Let me check the knowledge base for you again. One moment! :books:
Based on the information you've provided and the knowledge sources available, it seems unlikely that the error you're experiencing is due to the 'tier' parameter, as you're correctly using the 'model' parameter in your WebSocket URL. However, the issue could indeed be related to rate limits. Let's explore this possibility:
-
Rate Limits:
According to the API Rate Limits documentation, Deepgram does have rate limits in place for various services. For the Nova model (which includes nova-3), the rate limit is typically 100 concurrent requests per project. -
Concurrent Connections:
It's important to note that even if you're not hitting the maximum number of concurrent requests, you might still encounter rate limit errors. As explained in a GitHub discussion about TTS concurrency limits, rate limits are applied across an entire Deepgram project, regardless of which API keys are used. -
Intermittent Rate Limiting:
There have been instances where users experienced rate limiting even when they were below the stated limits. For example, in a GitHub discussion about response 429 errors, a user encountered rate limiting despite being well below the concurrent request limit. In that case, it was due to a temporary misconfiguration that has since been resolved.
To address potential rate limiting issues:
- Implement a backoff strategy:
As suggested in the Errors documentation, implement an exponential backoff retry strategy to handle 429 (Too Many Requests) errors. This can help manage high volumes of concurrent requests.
-
Monitor your usage:
Keep track of your concurrent connections and overall usage to ensure you're staying within the limits for your plan. -
Check for 429 errors:
If you're experiencing rate limiting, you should receive a 429 error with a message like this:{ "err_code": "TOO_MANY_REQUESTS", "err_msg": "Too many requests. Please try again later", "request_id": "uuid" } -
Consider upgrading your plan:
If you consistently need more concurrent connections, you might want to look into upgrading your plan. For example, the Streaming Text-to-Speech documentation mentions different concurrent connection limits for Pay As You Go (40) and Growth plans (80).