#How can I use my API key in Superwhisper for Mac?

1 messages · Page 1 of 1 (latest)

charred gale
#

I am hoping to connect Superwhisper for Mac to my Deepgram account so when I use the Nova Medical model I will have faster (?) service?

stray sinewBOT
#

Hi @charred gale. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

I'm sorry, but the provided knowledge sources do not contain information about how to configure your Deepgram API key specifically within the Superwhisper for Mac application.

However, the documentation does provide details on how to authenticate with Deepgram and use the Nova-2 Medical and Nova-3 Medical models, which you may need for your configuration:

  • API Key Authentication: Deepgram uses an Authorization header with the format Token <YOUR_DEEPGRAM_API_KEY> for its REST and WebSocket endpoints [Authentication; Pre-Recorded Authentication].
  • Medical Model Names:
    • To use the latest medical model, the parameter is model=nova-3-medical [Nova-3 Options].
    • To use the Nova-2 version, the parameter is model=nova-2-medical [Nova-2 Options].
  • Performance: Nova-3 is Deepgram's highest-performing general-purpose model, offering significant reductions in word error rate (WER) for both streaming and batch processing compared to previous models [Nova-3 Overview].

If Superwhisper requires a custom endpoint URL, the standard Deepgram transcription endpoint is <https://api.deepgram.com/v1/listen> [Pre-Recorded Audio]. I recommend checking the Superwhisper documentation or settings menu for fields labeled "API Key," "Model name," or "Custom Endpoint" to input these values.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@charred gale Kapa.ai is still learning and improving, please let me know how I did by reacting below.

charred gale
#

@stray sinew There is an option to set up a "Custom" model with an apixaban key. It has to be an "OpenAI compatible provider"...Will deepgram work?

stray sinewBOT
#

Let me check the knowledge base for you again. One moment! :books:

#

The provided knowledge sources do not explicitly mention if Deepgram's standalone Speech-to-Text (STT) or Text-to-Speech (TTS) APIs are fully "OpenAI compatible" in a way that allows them to be swapped into a generic OpenAI-formatted field for transcription.

However, the knowledge sources do mention the following regarding OpenAI compatibility:

  • Voice Agent API: Deepgram's Voice Agent API supports LLM providers that follow the OpenAI Chat Completions format. For example, setting provider.type to open_ai can be used with any endpoint that behaves like OpenAI’s Chat Completion API [LLM Models].
  • TTS Integration: Within the Voice Agent configuration, you can define multiple TTS providers, including OpenAI. Deepgram allows you to specify an endpoint with a url and headers (like an authorization Bearer token) for these external providers [Using Multiple TTS Providers].
  • Migration Differences: In a comparison between Deepgram and OpenAI, the documentation notes significant differences in features and response formats. For instance, Deepgram provides json, srt, vtt, and text formats, while OpenAI's transcription formats include verbose_json [Migrating From OpenAI Whisper to Deepgram].
#

If the "Custom" model option in your software expects an OpenAI-compatible LLM endpoint, Deepgram's Voice Agent API utilizes that format for its "Think" (LLM) step [Passing a custom (BYO) LLM through a Cloud Provider]. If it expects an OpenAI-compatible Transcription (Whisper) endpoint, the knowledge sources do not confirm that Deepgram's STT API follows the exact OpenAI request/response schema.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon: