feat(stage-ui): add GPT-SoVITS speech provider support

apps/stage-tamagotchi/src/main/index.ts

@@ -28,6 +28,7 @@ import { setupServerChannel } from './services/airi/channel-server'
 import { setupMcpStdioManager } from './services/airi/mcp-servers'
 import { setupPluginHost } from './services/airi/plugins'
 import { setupAutoUpdater } from './services/electron/auto-updater'
+import { createFetchService } from './services/electron/fetch'
 import { setupTray } from './tray'
 import { setupAboutWindowReusable } from './windows/about'
 import { setupBeatSync } from './windows/beat-sync'
@@ -97,6 +98,8 @@ app.whenReady().then(async () => {
     void fileLogger.appendLog(formatted)
   })
 
+  createFetchService()
+
   injeca.setLogger(createLoggLogger(useLogg('injeca').useGlobalConfig()))
 
   const appConfig = injeca.provide('configs:app', () => createGlobalAppConfig())

apps/stage-tamagotchi/src/main/services/electron/fetch.ts

@@ -0,0 +1,65 @@
+import { ipcMain } from 'electron'
+
+export const ELECTRON_FETCH_IPC_CHANNEL = 'airi:electron:fetch'
+
+export interface ElectronFetchPayload {
+  url: string
+  method?: string
+  headers?: Record<string, string>
+  body?: string
+}
+
+export interface ElectronFetchResult {
+  ok: boolean
+  status: number
+  statusText: string
+  headers: Record<string, string>
+  body: string
+  bodyBase64?: string
+}
+
+let fetchServiceRegistered = false
+
+export function createFetchService() {
+  if (fetchServiceRegistered)
+    return
+
+  fetchServiceRegistered = true
+
+  ipcMain.handle(ELECTRON_FETCH_IPC_CHANNEL, async (_event, payload: ElectronFetchPayload): Promise<ElectronFetchResult> => {
+    const response = await fetch(payload.url, {
+      method: payload.method ?? 'GET',
+      headers: payload.headers,
+      body: payload.body,
+    })
+
+    const headers: Record<string, string> = {}
+    response.headers.forEach((value, key) => {
+      headers[key] = value
+    })
+
+    const contentType = headers['content-type'] || ''
+    const isBinary = contentType.includes('audio/') || contentType.includes('image/') || contentType.includes('video/') || contentType.includes('application/octet-stream')
+
+    if (isBinary) {
+      const arrayBuffer = await response.arrayBuffer()
+      const base64 = Buffer.from(arrayBuffer).toString('base64')
+      return {
+        ok: response.ok,
+        status: response.status,
+        statusText: response.statusText,
+        headers,
+        body: '',
+        bodyBase64: base64,
+      }
+    }
+
+    return {
+      ok: response.ok,
+      status: response.status,
+      statusText: response.statusText,
+      headers,
+      body: await response.text(),
+    }
+  })
+}

apps/stage-tamagotchi/src/main/windows/settings/rpc/index.electron.ts

@@ -16,7 +16,7 @@ import { electronOpenDevtoolsWindow, electronOpenSettingsDevtools } from '../../
 import { createAuthService } from '../../../services/airi/auth'
 import { createMcpServersService } from '../../../services/airi/mcp-servers'
 import { createWidgetsService } from '../../../services/airi/widgets'
-import { createAutoUpdaterService } from '../../../services/electron'
+import { createAutoUpdaterService } from '../../../services/electron/auto-updater'
 import { setupBaseWindowElectronInvokes } from '../../shared/window'
 
 export async function setupSettingsWindowInvokes(params: {

nix/assets-hash.txt

@@ -1 +0,0 @@
-sha256-69tCpJaxRUnyR9CrHlmWJWEWLDpYzyzska7ui9++QoY=

packages/i18n/src/locales/en/settings.yaml

@@ -820,6 +820,34 @@ pages:
               description: Speech Service region
               label: Region
         title: Microsoft / Azure Speech
+      gpt-sovits:
+        description: github.com/RVC-Boss/GPT-SoVITS
+        title: GPT-SoVITS
+        callout_no_model_title: No model field support
+        callout_model_switch: GPT-SoVITS does not use the standard TTS model field. To switch models, fill in the GPT / SoVITS weight paths below and AIRI will call the api_v2.py switching endpoints.
+        callout_no_api_key: GPT-SoVITS does not require an API key. The field above can be left blank.
+        fields:
+          baseUrl:
+            description: Base URL of the GPT-SoVITS api_v2.py server
+            label: Base URL
+          refAudioPath:
+            description: Absolute path to reference audio file on the server
+            label: Reference Audio Path
+          promptText:
+            description: The text spoken in the reference audio, optional
+            label: Reference Audio Text
+          promptLang:
+            description: Language of the reference audio (zh / en / ja / ko / yue)
+            label: Reference Audio Language
+          textLang:
+            description: Language of the text to synthesize (zh / en / ja / ko / yue)
+            label: Synthesis Language
+          gptWeightsPath:
+            description: GPT weights path. If filled, AIRI will call /set_gpt_weights.
+            label: GPT Weights Path
+          sovitsWeightsPath:
+            description: SoVITS weights path. If filled, AIRI will call /set_sovits_weights.
+            label: SoVITS Weights Path
       index-tts-vllm:
         description: https://index-tts.github.io/
         title: Bilibili / IndexTTS

packages/i18n/src/locales/zh-Hans/settings.yaml

@@ -785,6 +785,34 @@ pages:
               description: 服务 Endpoint 地区（比如亚太 eastasia 区域）
               label: Endpoint 地区
         title: Microsoft / Azure 语音服务
+      gpt-sovits:
+        description: github.com/RVC-Boss/GPT-SoVITS
+        title: GPT-SoVITS
+        callout_no_model_title: 不支持请求内模型字段
+        callout_model_switch: GPT-SoVITS 不使用标准 TTS 的 model 字段；如需切换模型，请填写下方 GPT / SoVITS 权重路径，AIRI 会调用 api_v2.py 的切换接口。
+        callout_no_api_key: GPT-SoVITS 不需要 API Key，上方字段留空即可。
+        fields:
+          baseUrl:
+            description: GPT-SoVITS api_v2.py 服务器的 Base URL
+            label: Base URL
+          refAudioPath:
+            description: 服务器上参考音频文件的绝对路径
+            label: 参考音频路径
+          promptText:
+            description: 参考音频中说的文字内容，可留空
+            label: 参考音频文本
+          promptLang:
+            description: 参考音频的语言（zh / en / ja / ko / yue）
+            label: 参考音频语言
+          textLang:
+            description: 需要合成的文本语言（zh / en / ja / ko / yue）
+            label: 合成语言
+          gptWeightsPath:
+            description: GPT 权重路径；填写后会调用 /set_gpt_weights 切换
+            label: GPT 权重路径
+          sovitsWeightsPath:
+            description: SoVITS 权重路径；填写后会调用 /set_sovits_weights 切换
+            label: SoVITS 权重路径
       index-tts-vllm:
         description: https://index-tts.github.io/
         title: Bilibili / IndexTTS

packages/stage-pages/src/pages/settings/providers/speech/gpt-sovits.vue

@@ -0,0 +1,175 @@
+<script setup lang="ts">
+import type { SpeechProvider } from '@xsai-ext/providers/utils'
+
+import {
+  SpeechPlayground,
+  SpeechProviderSettings,
+} from '@proj-airi/stage-ui/components'
+import { useSpeechStore } from '@proj-airi/stage-ui/stores/modules/speech'
+import { useProvidersStore } from '@proj-airi/stage-ui/stores/providers'
+import { FieldInput } from '@proj-airi/ui'
+import { storeToRefs } from 'pinia'
+import { computed, onMounted, watch } from 'vue'
+import { useI18n } from 'vue-i18n'
+
+const { t } = useI18n()
+
+const providerId = 'gpt-sovits'
+const defaultModel = 'gpt-sovits'
+
+const speechStore = useSpeechStore()
+const providersStore = useProvidersStore()
+const { providers } = storeToRefs(providersStore)
+
+const refAudioPath = computed({
+  get: () => providers.value[providerId]?.refAudioPath as string | undefined || '',
+  set: (value) => {
+    if (!providers.value[providerId])
+      providers.value[providerId] = {}
+    providers.value[providerId].refAudioPath = value
+  },
+})
+
+const promptText = computed({
+  get: () => providers.value[providerId]?.promptText as string | undefined || '',
+  set: (value) => {
+    if (!providers.value[providerId])
+      providers.value[providerId] = {}
+    providers.value[providerId].promptText = value
+  },
+})
+
+const promptLang = computed({
+  get: () => providers.value[providerId]?.promptLang as string | undefined || 'zh',
+  set: (value) => {
+    if (!providers.value[providerId])
+      providers.value[providerId] = {}
+    providers.value[providerId].promptLang = value
+  },
+})
+
+const textLang = computed({
+  get: () => providers.value[providerId]?.textLang as string | undefined || 'zh',
+  set: (value) => {
+    if (!providers.value[providerId])
+      providers.value[providerId] = {}
+    providers.value[providerId].textLang = value
+  },
+})
+
+const gptWeightsPath = computed({
+  get: () => providers.value[providerId]?.gptWeightsPath as string | undefined || '',
+  set: (value) => {
+    if (!providers.value[providerId])
+      providers.value[providerId] = {}
+    providers.value[providerId].gptWeightsPath = value
+  },
+})
+
+const sovitsWeightsPath = computed({
+  get: () => providers.value[providerId]?.sovitsWeightsPath as string | undefined || '',
+  set: (value) => {
+    if (!providers.value[providerId])
+      providers.value[providerId] = {}
+    providers.value[providerId].sovitsWeightsPath = value
+  },
+})
+
+const apiKeyConfigured = computed(() => !!providers.value[providerId]?.baseUrl)
+
+const availableVoices = computed(() => speechStore.availableVoices[providerId] || [])
+
+onMounted(async () => {
+  await speechStore.loadVoicesForProvider(providerId)
+})
+
+watch([apiKeyConfigured], async () => {
+  await speechStore.loadVoicesForProvider(providerId)
+})
+
+async function handleGenerateSpeech(input: string, voiceId: string) {
+  const provider = await providersStore.getProviderInstance(providerId) as SpeechProvider
+  if (!provider)
+    throw new Error('Failed to initialize speech provider')
+
+  const providerConfig = providersStore.getProviderConfig(providerId)
+
+  return await speechStore.speech(provider, defaultModel, input, voiceId, providerConfig)
+}
+</script>
+
+<template>
+  <SpeechProviderSettings :provider-id="providerId" :default-model="defaultModel">
+    <template #basic-settings>
+      <p :class="['text-sm', 'text-blue-500', 'dark:text-blue-400', 'px-1']">
+        ⓘ {{ t('settings.pages.providers.provider.gpt-sovits.callout_model_switch') }}
+      </p>
+      <p :class="['text-sm', 'text-neutral-400', 'dark:text-neutral-500', 'px-1']">
+        ⓘ {{ t('settings.pages.providers.provider.gpt-sovits.callout_no_api_key') }}
+      </p>
+      <FieldInput
+        v-model="refAudioPath"
+        :label="t('settings.pages.providers.provider.gpt-sovits.fields.refAudioPath.label')"
+        :description="t('settings.pages.providers.provider.gpt-sovits.fields.refAudioPath.description')"
+        placeholder="/path/to/reference.wav"
+        required
+        type="text"
+      />
+      <FieldInput
+        v-model="promptText"
+        :label="t('settings.pages.providers.provider.gpt-sovits.fields.promptText.label')"
+        :description="t('settings.pages.providers.provider.gpt-sovits.fields.promptText.description')"
+        placeholder="参考音频中说的文字（可选）"
+        type="text"
+      />
+      <FieldInput
+        v-model="gptWeightsPath"
+        :label="t('settings.pages.providers.provider.gpt-sovits.fields.gptWeightsPath.label')"
+        :description="t('settings.pages.providers.provider.gpt-sovits.fields.gptWeightsPath.description')"
+        placeholder="/path/to/gpt.ckpt"
+        type="text"
+      />
+      <FieldInput
+        v-model="sovitsWeightsPath"
+        :label="t('settings.pages.providers.provider.gpt-sovits.fields.sovitsWeightsPath.label')"
+        :description="t('settings.pages.providers.provider.gpt-sovits.fields.sovitsWeightsPath.description')"
+        placeholder="/path/to/sovits.pth"
+        type="text"
+      />
+    </template>
+
+    <template #voice-settings>
+      <FieldInput
+        v-model="promptLang"
+        :label="t('settings.pages.providers.provider.gpt-sovits.fields.promptLang.label')"
+        :description="t('settings.pages.providers.provider.gpt-sovits.fields.promptLang.description')"
+        placeholder="zh"
+        type="text"
+      />
+      <FieldInput
+        v-model="textLang"
+        :label="t('settings.pages.providers.provider.gpt-sovits.fields.textLang.label')"
+        :description="t('settings.pages.providers.provider.gpt-sovits.fields.textLang.description')"
+        placeholder="zh"
+        type="text"
+      />
+    </template>
+
+    <template #playground>
+      <SpeechPlayground
+        :available-voices="availableVoices"
+        :generate-speech="handleGenerateSpeech"
+        :api-key-configured="apiKeyConfigured"
+        :use-ssml="false"
+        default-text="你好，我是 AIRI，很高兴认识你！"
+      />
+    </template>
+  </SpeechProviderSettings>
+</template>
+
+<route lang="yaml">
+  meta:
+    layout: settings
+    stageTransition:
+      name: slide
+  </route>