Summary
The Braintrust Java SDK does not instrument the official Deepgram Java SDK (com.deepgram:deepgram-java-sdk). Deepgram is a speech-AI provider whose Java SDK exposes execution-oriented generative/audio-AI surfaces — text-to-speech synthesis (Aura models), speech-to-text transcription, and a real-time Voice Agent API that orchestrates STT → LLM → TTS over a single WebSocket connection, including client/server function-call execution. None of this is instrumented anywhere in this repository, and there is no reference to Deepgram at all in the codebase.
What is missing
The com.deepgram:deepgram-java-sdk SDK (official, Fern-generated from Deepgram's OpenAPI spec) provides these execution surfaces, none of which are instrumented:
| API surface |
Description |
Text-to-speech (speak) |
Generative audio synthesis via Deepgram's Aura TTS models |
Speech-to-text (listen / transcription) |
Prerecorded and real-time (streaming) transcription |
Voice Agent API (agent().v1()...) |
Real-time voice-agent execution over WebSocket: STT + LLM "think" step + TTS orchestration, including client_side/server-side function-call execution during a live agent session |
There is no instrumentation module, wrapper, test, example, or dependency for Deepgram anywhere in this repository — a full-repo, case-insensitive search for deepgram returns zero matches.
Braintrust docs status
- Checked https://www.braintrust.dev/docs/integrations — Deepgram is not listed among the documented AI providers (which include Anthropic, OpenAI, Gemini, AWS Bedrock, Cohere, Mistral, Groq, xAI, and others) for any language: not_found
Weekly downloads
Weekly downloads: 775,998 (as of 2026-08-17; https://pypistats.org/packages/deepgram-sdk — official Python SDK, the ecosystem's primary tracked distribution; Maven Central does not publish download counts for the Java artifact com.deepgram:deepgram-java-sdk). For cross-reference, the official JS/TS SDK @deepgram/sdk recorded 537,899 downloads for the week of 2026-08-09–2026-08-15 (https://api.npmjs.org/downloads/point/last-week/@deepgram/sdk).
Upstream sources
Local files inspected
- Full-repo case-insensitive grep for
deepgram across all files — zero matches
braintrust-sdk/instrumentation/ — contains openai_2_15_0, anthropic_2_2_0, genai_1_18_0, langchain_1_8_0, springai_1_0_0, springai_2_0_0, aws_bedrock_2_30_0; no Deepgram directoryREADME.md — lists OpenAI, Anthropic, and other LLM providers as supported; no mention of Deepgram or any speech/voice provider
Summary
The Braintrust Java SDK does not instrument the official Deepgram Java SDK (
com.deepgram:deepgram-java-sdk). Deepgram is a speech-AI provider whose Java SDK exposes execution-oriented generative/audio-AI surfaces — text-to-speech synthesis (Aura models), speech-to-text transcription, and a real-time Voice Agent API that orchestrates STT → LLM → TTS over a single WebSocket connection, including client/server function-call execution. None of this is instrumented anywhere in this repository, and there is no reference to Deepgram at all in the codebase.What is missing
The
com.deepgram:deepgram-java-sdkSDK (official, Fern-generated from Deepgram's OpenAPI spec) provides these execution surfaces, none of which are instrumented:speak)listen/ transcription)agent().v1()...)client_side/server-side function-call execution during a live agent sessionThere is no instrumentation module, wrapper, test, example, or dependency for Deepgram anywhere in this repository — a full-repo, case-insensitive search for
deepgramreturns zero matches.Braintrust docs status
Weekly downloads
Weekly downloads: 775,998 (as of 2026-08-17; https://pypistats.org/packages/deepgram-sdk — official Python SDK, the ecosystem's primary tracked distribution; Maven Central does not publish download counts for the Java artifact
com.deepgram:deepgram-java-sdk). For cross-reference, the official JS/TS SDK@deepgram/sdkrecorded 537,899 downloads for the week of 2026-08-09–2026-08-15 (https://api.npmjs.org/downloads/point/last-week/@deepgram/sdk).Upstream sources
Local files inspected
deepgramacross all files — zero matchesbraintrust-sdk/instrumentation/— containsopenai_2_15_0,anthropic_2_2_0,genai_1_18_0,langchain_1_8_0,springai_1_0_0,springai_2_0_0,aws_bedrock_2_30_0; no Deepgram directoryREADME.md— lists OpenAI, Anthropic, and other LLM providers as supported; no mention of Deepgram or any speech/voice provider