feat(voice): 语音配置控制面——火山豆包 appid/token/resource-id/音色(参数落地处)

JARVIS 不依赖火山账号能先做的第二块:配置层,做成'你参数一填就落库'的形态。
- voice.Config: AppID/AccessToken/ASRResourceID/TTSResourceID/TTSVoiceType,AccessToken
  走 secrets AES 加密入库(镜像微信配置);ASREnabled/TTSEnabled 可各自独立判断。
- admin GET/PUT /admin/voice(RequireAdmin),token 空串=沿用已存、明文回显。
- admin「语音设置」页(运维组):五个字段 + ASR/TTS 就绪徽标 + '去哪拿参数'清单。

端点/音频格式(PCM 16k 单声道)由代码固定不入配置。WS 会话 + 火山 ASR/TTS 客户端等
用户回传 resource-id 后按官方 demo 协议做。build+tsc 绿。

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
Blizzard
2026-07-21 17:00:35 +08:00
parent 644b635d73
commit 618184689e
6 changed files with 295 additions and 0 deletions
+32
View File
@@ -267,6 +267,38 @@ export async function saveWechatMP(body: { appid: string; app_secret: string; to
return { enabled: !!d.enabled };
}
// —— 语音(火山豆包)配置 ——
export interface VoiceConfig {
appid: string;
access_token: string; // 明文回显(单管理员后台)
asr_resource_id: string;
tts_resource_id: string;
tts_voice_type: string;
asr_enabled: boolean;
tts_enabled: boolean;
}
export async function getVoiceConfig(): Promise<VoiceConfig> {
const res = guard(await fetch(`${ADMIN}/voice`, { headers: authHeaders() }));
const d = (await res.json().catch(() => ({}))) as Partial<VoiceConfig> & { error?: string };
if (!res.ok) throw new Error(d.error ?? `voice config failed: ${res.status}`);
return {
appid: d.appid ?? "", access_token: d.access_token ?? "", asr_resource_id: d.asr_resource_id ?? "",
tts_resource_id: d.tts_resource_id ?? "", tts_voice_type: d.tts_voice_type ?? "",
asr_enabled: !!d.asr_enabled, tts_enabled: !!d.tts_enabled,
};
}
// access_token 传空串 = 沿用已保存的。
export async function saveVoiceConfig(body: {
appid: string; access_token: string; asr_resource_id: string; tts_resource_id: string; tts_voice_type: string;
}): Promise<{ asr_enabled: boolean; tts_enabled: boolean }> {
const res = guard(await fetch(`${ADMIN}/voice`, { method: "PUT", headers: authHeaders(true), body: JSON.stringify(body) }));
const d = (await res.json().catch(() => ({}))) as { asr_enabled?: boolean; tts_enabled?: boolean; error?: string };
if (!res.ok) throw new Error(d.error ?? `save failed: ${res.status}`);
return { asr_enabled: !!d.asr_enabled, tts_enabled: !!d.tts_enabled };
}
export interface WechatPayConfig {
mchid: string;
cert_serial: string;
@@ -0,0 +1,122 @@
import { useEffect, useState } from "react";
import { getVoiceConfig, saveVoiceConfig } from "../api";
// 运维 · 语音设置:火山引擎豆包语音(流式 ASR 耳朵 + 双向流式 TTS 嘴)。
// AccessToken 明文回显(单管理员后台,方便核对;密文仍加密入库)。设计见 VOICE_DESIGN.md。
export function VoiceConfigPage() {
const [appid, setAppid] = useState("");
const [token, setToken] = useState("");
const [asrRes, setAsrRes] = useState("");
const [ttsRes, setTtsRes] = useState("");
const [voice, setVoice] = useState("");
const [asrOn, setAsrOn] = useState(false);
const [ttsOn, setTtsOn] = useState(false);
const [busy, setBusy] = useState(false);
const [err, setErr] = useState("");
const [ok, setOk] = useState(false);
useEffect(() => {
getVoiceConfig()
.then((c) => {
setAppid(c.appid);
setToken(c.access_token);
setAsrRes(c.asr_resource_id);
setTtsRes(c.tts_resource_id);
setVoice(c.tts_voice_type);
setAsrOn(c.asr_enabled);
setTtsOn(c.tts_enabled);
})
.catch((e) => setErr((e as Error).message));
}, []);
const save = async () => {
if (busy) return;
setBusy(true);
setErr("");
setOk(false);
try {
const r = await saveVoiceConfig({
appid: appid.trim(),
access_token: token,
asr_resource_id: asrRes.trim(),
tts_resource_id: ttsRes.trim(),
tts_voice_type: voice.trim(),
});
setAsrOn(r.asr_enabled);
setTtsOn(r.tts_enabled);
setOk(true);
} catch (e) {
setErr((e as Error).message);
} finally {
setBusy(false);
}
};
const field = "mt-1 block w-full rounded-lg border border-gray-200 px-2.5 py-1.5 font-mono text-sm text-gray-800 focus:border-violet-400 focus:outline-none";
const badge = (on: boolean, label: string) =>
on ? (
<span className="rounded bg-emerald-50 px-2 py-0.5 text-[10px] font-medium text-emerald-600">{label} </span>
) : (
<span className="rounded bg-gray-100 px-2 py-0.5 text-[10px] text-gray-500">{label} </span>
);
return (
<div className="space-y-5">
<div className="border-b border-gray-200 pb-4">
<h3 className="text-base font-semibold text-gray-800"> · </h3>
<p className="text-xs text-gray-400">() + ()AccessToken </p>
</div>
<div className="rounded-xl border border-gray-100 bg-white p-5 shadow-sm">
<div className="mb-3 flex items-center gap-2">
<h4 className="text-sm font-semibold text-gray-700"></h4>
{badge(asrOn, "ASR")}
{badge(ttsOn, "TTS")}
</div>
<div className="grid grid-cols-1 gap-3 md:grid-cols-2">
<label className="text-xs text-gray-500">
AppID(X-Api-App-Key)
<input value={appid} onChange={(e) => setAppid(e.target.value)} placeholder="火山应用 AppID" className={field} />
</label>
<label className="text-xs text-gray-500">
Access Token(X-Api-Access-Key)
<input value={token} onChange={(e) => setToken(e.target.value)} placeholder="火山 Access Token" className={field} />
</label>
<label className="text-xs text-gray-500">
ASR Resource-Id
<input value={asrRes} onChange={(e) => setAsrRes(e.target.value)} placeholder="流式语音识别的 X-Api-Resource-Id" className={field} />
</label>
<label className="text-xs text-gray-500">
TTS Resource-Id
<input value={ttsRes} onChange={(e) => setTtsRes(e.target.value)} placeholder="双向流式 TTS 的 X-Api-Resource-Id" className={field} />
</label>
<label className="text-xs text-gray-500 md:col-span-2">
voice_type
<input value={voice} onChange={(e) => setVoice(e.target.value)} placeholder="从标准音色里挑一个(如 zh_male_…)" className={field} />
</label>
</div>
{err && <p className="mt-2 text-xs text-rose-500">{err}</p>}
{ok && <p className="mt-2 text-xs text-emerald-600"></p>}
<div className="mt-3">
<button onClick={() => void save()} disabled={busy} className="rounded-lg bg-violet-600 px-3.5 py-1.5 text-sm text-white hover:bg-violet-700 disabled:opacity-40">
{busy ? "保存中…" : "保存"}
</button>
</div>
</div>
<div className="rounded-xl border border-amber-100 bg-amber-50/40 p-5 text-xs leading-relaxed text-gray-600">
<h4 className="mb-2 text-sm font-semibold text-gray-700"></h4>
<ol className="list-decimal space-y-1.5 pl-4">
<li> <b>AppID</b> <b>Access Token</b>ASR/TTS </li>
<li><b>ASR Resource-Id</b> 2.0X-Api-Resource-Id </li>
<li><b>TTS Resource-Id</b></li>
<li><b></b>JARVIS voice_type</li>
<li>/PCM 16k </li>
</ol>
</div>
</div>
);
}
+8
View File
@@ -1,5 +1,6 @@
import { SubscriptionPage } from "./pages/SubscriptionPage";
import { LoginConfigPage } from "./pages/LoginConfigPage";
import { VoiceConfigPage } from "./pages/VoiceConfigPage";
import { WechatUsersPage } from "./pages/WechatUsersPage";
import { lazy, type ReactNode } from "react";
@@ -86,6 +87,13 @@ export const routes: RouteDef[] = [
ready: true,
element: <LoginConfigPage />,
},
{
path: "voice-config",
label: "语音设置",
group: "运维",
ready: true,
element: <VoiceConfigPage />,
},
{
path: "datasources",
label: "数据源 & RAG",
@@ -0,0 +1,81 @@
package handler
import (
"context"
"encoding/json"
"net/http"
"strings"
"github.com/gin-gonic/gin"
"github.com/sundynix/sundynix-gateway/internal/voice"
)
// 语音(火山引擎豆包语音)配置的管理端存取。设计见 VOICE_DESIGN.md。
// AccessToken AES 加密入库;单管理员后台明文回显(同微信配置),方便核对/复制。
const SettingVoice = "voice_config" // 语音配置(setting 表)
func (h *Handler) loadVoiceConfig(ctx context.Context) voice.Config {
raw := h.db.GetSetting(ctx, SettingVoice)
if raw == "" {
return voice.Config{}
}
var c voice.Config
if json.Unmarshal([]byte(raw), &c) != nil {
return voice.Config{}
}
return c.DecryptFromStore()
}
// AdminGetVoiceConfig: GET /api/v1/admin/voice —— 回显语音配置(token 明文,RequireAdmin 已拦)。
func (h *Handler) AdminGetVoiceConfig(c *gin.Context) {
cfg := h.loadVoiceConfig(c.Request.Context())
c.JSON(http.StatusOK, gin.H{
"appid": cfg.AppID,
"access_token": cfg.AccessToken,
"asr_resource_id": cfg.ASRResourceID,
"tts_resource_id": cfg.TTSResourceID,
"tts_voice_type": cfg.TTSVoiceType,
"asr_enabled": cfg.ASREnabled(),
"tts_enabled": cfg.TTSEnabled(),
})
}
// AdminSaveVoiceConfig: PUT /api/v1/admin/voice —— 保存语音配置(token 空串=沿用已存)。
func (h *Handler) AdminSaveVoiceConfig(c *gin.Context) {
var b struct {
AppID string `json:"appid"`
AccessToken string `json:"access_token"`
ASRResourceID string `json:"asr_resource_id"`
TTSResourceID string `json:"tts_resource_id"`
TTSVoiceType string `json:"tts_voice_type"`
}
if err := c.ShouldBindJSON(&b); err != nil {
c.JSON(http.StatusBadRequest, gin.H{"error": "参数错误"})
return
}
ctx := c.Request.Context()
token := strings.TrimSpace(b.AccessToken)
if token == "" {
token = h.loadVoiceConfig(ctx).AccessToken // 留空=沿用已存
}
cfg := voice.Config{
AppID: strings.TrimSpace(b.AppID),
AccessToken: token,
ASRResourceID: strings.TrimSpace(b.ASRResourceID),
TTSResourceID: strings.TrimSpace(b.TTSResourceID),
TTSVoiceType: strings.TrimSpace(b.TTSVoiceType),
}
stored, err := cfg.EncryptedForStore()
if err != nil {
c.JSON(http.StatusInternalServerError, gin.H{"error": "密钥加密失败: " + err.Error()})
return
}
raw, _ := json.Marshal(stored)
if err := h.db.SetSetting(ctx, SettingVoice, string(raw)); err != nil {
c.JSON(http.StatusBadGateway, gin.H{"error": err.Error()})
return
}
c.JSON(http.StatusOK, gin.H{"status": "ok", "asr_enabled": cfg.ASREnabled(), "tts_enabled": cfg.TTSEnabled()})
}
@@ -169,6 +169,8 @@ func New(db *store.Postgres, cache *store.Redis, bus *nats.Bus, blobStore *blob.
admin.GET("/wechat-users", h.AdminWechatUsers) // 微信用户列表(openid/加入时间/余额)
admin.GET("/payment/wechat", h.AdminGetWechatPay)
admin.PUT("/payment/wechat", h.AdminSaveWechatPay)
admin.GET("/voice", h.AdminGetVoiceConfig) // 语音(火山豆包)配置
admin.PUT("/voice", h.AdminSaveVoiceConfig)
admin.GET("/sub-plans", h.AdminSubPlans) // 订阅套餐(含下架)
admin.PUT("/sub-plans", h.AdminSaveSubPlan) // 配价格/时长/发放节奏
admin.GET("/subscriptions", h.AdminSubscriptions) // 全平台订阅观测
+50
View File
@@ -0,0 +1,50 @@
package voice
import "github.com/sundynix/sundynix-shared/secrets"
// Config 是语音交互(火山引擎豆包语音)所需配置。AccessToken 加密入库、后台明文回显(同微信配置)。
//
// 火山鉴权走 WS 握手 headerX-Api-App-Key(AppID) / X-Api-Access-Key(AccessToken) /
// X-Api-Resource-Id(区分服务,ASR 与双向 TTS 各一个)。AppID/Token 通常同一应用共用。
// 端点/音频格式(PCM 16k 单声道等)由代码固定,不入用户配置。
type Config struct {
AppID string `json:"appid"` // 火山应用 AppIDX-Api-App-Key
AccessToken string `json:"access_token"` // 火山 Access TokenX-Api-Access-Key,密文入库)
ASRResourceID string `json:"asr_resource_id"` // 流式语音识别 X-Api-Resource-Id
TTSResourceID string `json:"tts_resource_id"` // 双向流式 TTS X-Api-Resource-Id
TTSVoiceType string `json:"tts_voice_type"` // 音色(如 zh_male_… / BV700_streaming
}
// ASREnabled / TTSEnabled 分别报告耳朵、嘴是否配齐(可各自独立启用)。
func (c Config) ASREnabled() bool {
return c.AppID != "" && c.AccessToken != "" && c.ASRResourceID != ""
}
func (c Config) TTSEnabled() bool {
return c.AppID != "" && c.AccessToken != "" && c.TTSResourceID != "" && c.TTSVoiceType != ""
}
// Enabled 报告语音整体是否可用(耳朵 + 嘴都配齐)。
func (c Config) Enabled() bool { return c.ASREnabled() && c.TTSEnabled() }
// EncryptedForStore 返回 AccessToken 已加密的副本,用于落库。
func (c Config) EncryptedForStore() (Config, error) {
if c.AccessToken == "" {
return c, nil
}
enc, err := secrets.Encrypt(c.AccessToken)
if err != nil {
return c, err
}
c.AccessToken = enc
return c, nil
}
// DecryptFromStore 把库内密文 AccessToken 还原为明文。
func (c Config) DecryptFromStore() Config {
if c.AccessToken != "" {
if plain, err := secrets.Decrypt(c.AccessToken); err == nil {
c.AccessToken = plain
}
}
return c
}