Audio Providers

Video Context MCP supports four audio providers for the transcribe_video tool and for audio-enhanced video analysis. See Model Pricing for a summary of free credits and costs.

Provider Comparison

Feature	Deepgram	AssemblyAI	Groq/Whisper	Gemini
Default model	nova-3	universal-3-5-pro	whisper-large-v3	gemini-3.8-flash
Price	Paid ($200 free credits)	Paid ($50 free credits)	Free tier	Free tier
Speaker diarization	✓	✓ (highest quality)	✗	✗
English translation	✗	✗	✓	✓
Multilingual transcription	✓	✓	✓	✓
Best for	Default — fast, accurate	High-quality diarization	Free / cost-conscious	Users already using Gemini

Deepgram — Default

Fast, accurate, and highly reliable. Offers $200 in free credits to new accounts. The default audio provider. Uses Deepgram Nova-3 by default (DEEPGRAM_MODEL=nova-2 rolls back).

Environment variable: DEEPGRAM_API_KEY
Model selection: DEEPGRAM_MODEL — nova-3 (default) or nova-2 (rollback)
Get API key →

AssemblyAI

Best-in-class speaker diarization. $50 free credits to new accounts. Use this when identifying who said what is important. Uses Universal-3.5 Pro with an automatic Universal-2 fallback for languages it does not cover.

Environment variable: ASSEMBLYAI_API_KEY
Model selection: ASSEMBLYAI_MODEL — comma-separated list (default universal-3-5-pro,universal-2)
Get API key →

Groq/Whisper

Free tier available. Based on OpenAI Whisper (Whisper large-v3). The only free provider that supports translation to English — use this when you need an English transcript of non-English audio.

Environment variable: GROQ_API_KEY
Model selection: GROQ_AUDIO_MODEL — default whisper-large-v3 (e.g. whisper-large-v3-turbo)
Get API key →

Gemini

Free tier available. Reuses the same key as the Gemini video provider. Also supports English translation. A good option for users who already have a Gemini key configured.

Environment variable: GEMINI_API_KEY (same key as video)

Fallback Chain

When a provider fails or its key is missing, the server automatically tries the next one in order. With the default (Deepgram):

Deepgram → AssemblyAI → Groq → Gemini

Change the Default Audio Provider

Set AUDIO_MCP_DEFAULT_PROVIDER in your MCP environment config:

"AUDIO_MCP_DEFAULT_PROVIDER": "assemblyai"

Valid values: deepgram, assemblyai, groq, gemini

Tip

To ensure the LLM respects your configured default provider, add the following to your Copilot instructions (.github/copilot-instructions.md) or AGENTS.md:

If AUDIO_MCP_DEFAULT_PROVIDER is set in the MCP config, respect that value as the default provider for audio tools. Do not specify a provider parameter unless the user explicitly requests one. :::

Audio-Enhanced Video Analysis

When the AUDIO_ENHANCE_VIDEO_ANALYSIS variable is set to auto (the default), the server automatically transcribes audio and injects the transcript into AI prompts for analyze_video and summarize_video. This produces richer answers by giving the AI both visual and spoken context.

Value	Behavior
auto (default)	Transcribe when an audio track is detected
true	Always attempt transcription
false	Disable audio enhancement entirely

Note

“Translation” here means converting non-English speech to English in the final output. It is different from multilingual transcription — Deepgram and AssemblyAI natively transcribe dozens of languages but always output in the source language. Use Groq or Gemini when you need English output from non-English audio.