feat: 语音流式输入管线 + VAD前端集成 + 插件-工具合并清理

- 前端: VAD语音检测(@ricky0123/vad-web) + useVoiceInput双模式(流式WS/REST) - Gateway: VoiceStreamManager代理WS流式STT到voice-service - Voice-service: DashScope REST → Realtime WS → Whisper三级引擎 + ffmpeg转码 - 共享模块: pkg/audio(音频转换) + pkg/dashscope(ASR REST客户端) - 清理: 移除旧plugin-manager和pkg/plugins，完成插件→工具合并 - 文档: 完善gateway-api.md和voice-service.md语音API文档 - 工具: scripts/voice/ 语音转换脚本集 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-06 11:50:40 +08:00
parent 258cf81b25
commit 6ef9e082a6
91 changed files with 4091 additions and 3929 deletions
@@ -0,0 +1,127 @@
+package dashscope
+
+import (
+	"bytes"
+	"context"
+	"encoding/base64"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"time"
+)
+
+// ---- 共享类型 ----
+
+// ASRRequest DashScope ASR REST API 请求体。
+type ASRRequest struct {
+	Model      string     `json:"model"`
+	Input      ASRInput   `json:"input"`
+	Parameters ASRParams  `json:"parameters"`
+}
+
+// ASRInput 音频输入。
+type ASRInput struct {
+	Audio string `json:"audio"`
+}
+
+// ASRParams 识别参数。
+type ASRParams struct {
+	Format     string `json:"format,omitempty"`
+	SampleRate int    `json:"sample_rate,omitempty"`
+	Language   string `json:"language,omitempty"`
+}
+
+// ASRResponse DashScope ASR REST API 响应体。
+type ASRResponse struct {
+	Output struct {
+		Text string `json:"text"`
+	} `json:"output"`
+	Usage struct {
+		TotalTokens int `json:"total_tokens"`
+	} `json:"usage"`
+	RequestID string `json:"request_id"`
+	Code      string `json:"code,omitempty"`
+	Message   string `json:"message,omitempty"`
+}
+
+// ---- 共享客户端 ----
+
+// RESTClient 封装 DashScope REST API 的 HTTP 通信。
+type RESTClient struct {
+	apiKey string
+	client *http.Client
+}
+
+// NewRESTClient 创建 REST 客户端。
+func NewRESTClient(apiKey string) *RESTClient {
+	return &RESTClient{
+		apiKey: apiKey,
+		client: &http.Client{Timeout: 60 * time.Second},
+	}
+}
+
+// IsAvailable 检查 API Key 是否已配置。
+func (c *RESTClient) IsAvailable() bool {
+	return c.apiKey != ""
+}
+
+// Transcribe 调用 DashScope ASR REST API 进行语音识别。
+// audioData 应为 PCM 16kHz mono 格式。
+func (c *RESTClient) Transcribe(ctx context.Context, model string, audioData []byte, format string, sampleRate int, language string) (string, error) {
+	if !c.IsAvailable() {
+		return "", fmt.Errorf("DashScope ASR API key not configured")
+	}
+	if language == "" || language == "auto" {
+		language = "zh"
+	}
+
+	audioB64 := base64.StdEncoding.EncodeToString(audioData)
+
+	reqBody := ASRRequest{
+		Model: model,
+		Input: ASRInput{
+			Audio: fmt.Sprintf("data:audio/%s;base64,%s", format, audioB64),
+		},
+		Parameters: ASRParams{
+			Format:     format,
+			SampleRate: sampleRate,
+			Language:   language,
+		},
+	}
+
+	bodyBytes, err := json.Marshal(reqBody)
+	if err != nil {
+		return "", fmt.Errorf("marshal ASR request: %w", err)
+	}
+
+	url := "https://dashscope.aliyuncs.com/api/v1/services/audio/asr/asr"
+	req, err := http.NewRequestWithContext(ctx, "POST", url, bytes.NewReader(bodyBytes))
+	if err != nil {
+		return "", fmt.Errorf("create ASR request: %w", err)
+	}
+	req.Header.Set("Content-Type", "application/json")
+	req.Header.Set("Authorization", "Bearer "+c.apiKey)
+
+	resp, err := c.client.Do(req)
+	if err != nil {
+		return "", fmt.Errorf("ASR request failed: %w", err)
+	}
+	defer resp.Body.Close()
+
+	respBytes, err := io.ReadAll(resp.Body)
+	if err != nil {
+		return "", fmt.Errorf("read ASR response: %w", err)
+	}
+
+	var asrResp ASRResponse
+	if err := json.Unmarshal(respBytes, &asrResp); err != nil {
+		return "", fmt.Errorf("parse ASR response: %w", err)
+	}
+
+	if asrResp.Code != "" && asrResp.Code != "0" {
+		return "", fmt.Errorf("ASR error: %s (code=%s)", asrResp.Message, asrResp.Code)
+	}
+
+	return asrResp.Output.Text, nil
+}