Update Vertex AI Text to Speech doc to show use of audio

[Sameerlite]

Relevant issues

Fixes #18178

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/litellm/ directory, Adding at least 1 test is a hard requirement - see details
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem

CI (LiteLLM team)

CI status guideline:

• 50-55 passing tests: main is stable with minor issues.
• 45-49 passing tests: acceptable but needs attention
• <= 40 passing tests: unstable; be careful with your merges and assess the risk.

• Branch creation CI run
  Link:

• CI run for the last commit
  Link:

• Merge / cherry-pick CI run
  Links:

Type

🆕 New Feature
🐛 Bug Fix
🧹 Refactoring
📖 Documentation
🚄 Infrastructure
✅ Test

Changes

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
litellm	Building	Preview, Comment	Feb 2, 2026 10:46am

[greptile-apps]

Greptile Overview

Greptile Summary

This PR updates the Vertex AI Text to Speech documentation to clarify that when using Gemini TTS models via LiteLLM Proxy, users must include allowed_openai_params to enable the audio and modalities parameters.

Changes

• Added note in the warning section explaining the requirement for allowed_openai_params when using LiteLLM Proxy
• Updated the curl example to include "allowed_openai_params": ["audio", "modalities"] in the request body
• Updated the OpenAI Python SDK example to include extra_body={"allowed_openai_params": ["audio", "modalities"]}

Context

This addresses issue #18178. The allowed_openai_params parameter is necessary because:

1. LiteLLM validates which OpenAI parameters are supported by each provider/model
2. By default, if a parameter is not explicitly marked as supported, LiteLLM will raise an UnsupportedParamsError
3. The allowed_openai_params field allows users to dynamically override this validation and explicitly permit certain parameters
4. For Gemini TTS models accessed through the proxy, the audio and modalities parameters need to be explicitly allowed to pass through

The documentation change is consistent with LiteLLM's existing pattern for handling provider-specific parameters through the proxy, as documented in /docs/completion/drop_params.md.

Confidence Score: 4/5

• This PR is safe to merge with low risk - it's a documentation-only change that clarifies usage
• Documentation change is accurate and helpful. The score is 4/5 rather than 5/5 because: (1) the PR lacks tests as noted in the pre-submission checklist, though tests may not be strictly necessary for documentation changes, and (2) it would be helpful to verify that the documented pattern actually works as described
• No files require special attention

Important Files Changed

Filename	Overview
docs/my-website/docs/providers/vertex_speech.md	Added allowed_openai_params requirement for audio parameters when using Gemini TTS via proxy

Sequence Diagram

sequenceDiagram
    participant User
    participant OpenAI_SDK as OpenAI SDK Client
    participant LiteLLM_Proxy as LiteLLM Proxy
    participant Vertex_AI as Vertex AI Gemini TTS
    
    User->>OpenAI_SDK: Create chat completion request
    Note over User,OpenAI_SDK: messages, modalities=["audio"]<br/>audio={"voice": "Kore", "format": "pcm16"}
    OpenAI_SDK->>LiteLLM_Proxy: POST /v1/chat/completions
    Note over OpenAI_SDK,LiteLLM_Proxy: extra_body={"allowed_openai_params": ["audio", "modalities"]}
    
    LiteLLM_Proxy->>LiteLLM_Proxy: Check allowed_openai_params
    Note over LiteLLM_Proxy: Validates "audio" and "modalities"<br/>are in allowed list
    
    LiteLLM_Proxy->>LiteLLM_Proxy: Transform audio params to speechConfig
    Note over LiteLLM_Proxy: Maps OpenAI audio format<br/>to Vertex AI speechConfig
    
    LiteLLM_Proxy->>Vertex_AI: Generate content with audio
    Note over LiteLLM_Proxy,Vertex_AI: speechConfig={voiceConfig={prebuiltVoiceConfig={voiceName: "Kore"}}}<br/>responseModalities=["AUDIO"]
    
    Vertex_AI-->>LiteLLM_Proxy: Audio response (pcm16)
    LiteLLM_Proxy-->>OpenAI_SDK: OpenAI-format response
    OpenAI_SDK-->>User: Audio content

Loading

[greptile-apps]

_{1 file reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}