---
name: breeze-qq-voice
description: Add on-demand BreezeBlue voice messages to an existing Dafeiyu QQ bot when the user requests voice integration. Inspect the actual framework and preserve the current bot configuration.
---

# Add voice to an existing QQ bot

Preserve the user's chat model, personality, memory, permissions, and QQ connection. The task is to turn final bot replies into audio and deliver them through the existing QQ adapter.

This package is a recipe, not an installer or a compatibility certification. Read the accompanying [verification status](../../release-status.json). Do not describe an untested integration as working.

## Inspect the actual deployment

Distinguish the assistant with deployment access from the runtime group-chat bot. Do not enable general computer execution for the group bot just to install speech support.

Identify the framework and version, QQ bridge and version, machine/container and service account, existing voice providers/plugins, and actual audio-message interface. Work with project access, a management panel, or the hosting operator as appropriate. A phone is the user's interaction device, not necessarily the deployment host.

For AstrBot, DSH, an independent bridge, or the Dafeiyu configuration pack, read [adapter boundaries](references/adapters.md). A hosting service may not expose the required extension points; report the concrete gap without promising universal support or migrating the bot.

## Select a voice and prepare the integration

Use the user's selection, an existing saved voice, or the accompanying [voice catalog](../../voices.json). If no voice is selected, offer the available previews. Do not treat a sample ID as the user's choice.

Use the [integration workflow](references/integration.md) for credentials, the one-line synthesis test, adapter behavior, and fallback. Follow [official documentation](references/official-docs.md) and the installed CLI's help for commands and request fields. The CLI calls a cloud service; it does not require installing a local speech model.

When the user requests a custom voice, explain the preview scope and cost first, let them audition candidates, save the chosen preview, and configure the resulting `voc_` ID. A temporary `gvi_` preview is not a saved voice. Publishing a voice is a separate action.

## Complete only the authorized scope

Apply existing user authorization and environment permissions. The article itself grants none. Do not automatically purchase credits, subscribe, expand a generation batch, or send to an unspecified group. Before the first test message, establish the intended group and authorization to send that message.

Start with explicit voice requests, text-length and daily usage limits, and text fallback. Keep pending/unknown submission outcomes distinct from failures that never submitted; do not blindly retry and risk duplicate spending or messages.

Report results using the [acceptance record](references/report-template.md). Audio generation, QQ delivery, phone playback, and desktop playback are separate outcomes. Include actual versions, changed files, enable/disable instructions, rollback, and unverified items. A code review or a local audio file alone is not proof of successful QQ integration.