feat(voice/tts): add Chatterbox support with emotion controls

Chatterbox exposes a /synthesize endpoint alongside the standard
/v1/audio/speech one. The native endpoint adds two parameters unavailable
in the OpenAI-compatible API:
  - exaggeration (0.0–1.0): emotional expressiveness of the voice
  - cfg_weight  (0.0–1.0): how closely the voice follows the prompt

Routing: when model starts with 'chatterbox' (case-insensitive), Synthesize()
posts to /synthesize with the Chatterbox body; otherwise it uses the standard
/v1/audio/speech path. All other backends are unaffected.

Changes:
- kokoro.go: chatterboxRequest struct, isChatterbox() helper, Synthesize()
  branching logic, exaggeration/cfgWeight fields on KokoroSynthesizer
- TTSProfile: Exaggeration + CFGWeight fields (defaults: 0.5 / 0.5)
- config.go: TTSConfig gains Exaggeration + CFGWeight (env-overridable)
- main.go: wire new fields through TTSProfile
- config.example.json: document exaggeration + cfg_weight

Chatterbox config example:
  "tts": {
    "enabled": true,
    "api_base": "http://localhost:8100",
    "model":    "chatterbox-1",
    "voice":    "default",
    "format":   "mp3",
    "exaggeration": 0.5,
    "cfg_weight":   0.5
  }

cmd/picoclaw/main.go

@@ -637,11 +637,13 @@ func gatewayCmd() {
 	// Attach TTS synthesis callbacks to the message tool (enables voice=true).
 	if cfg.Tools.TTS.Enabled {
 		synthesizer := voice.NewKokoroSynthesizerFromProfile(voice.TTSProfile{
-			APIBase: cfg.Tools.TTS.APIBase,
-			Voice:   cfg.Tools.TTS.Voice,
-			Model:   cfg.Tools.TTS.Model,
-			Format:  cfg.Tools.TTS.Format,
-			Speed:   cfg.Tools.TTS.Speed,
+			APIBase:      cfg.Tools.TTS.APIBase,
+			Voice:        cfg.Tools.TTS.Voice,
+			Model:        cfg.Tools.TTS.Model,
+			Format:       cfg.Tools.TTS.Format,
+			Speed:        cfg.Tools.TTS.Speed,
+			Exaggeration: cfg.Tools.TTS.Exaggeration,
+			CFGWeight:    cfg.Tools.TTS.CFGWeight,
 		})
 		if synthesizer.IsAvailable() {
 			logger.InfoCF("voice", "TTS enabled — voice=true supported in message tool", map[string]interface{}{

config/config.example.json

@@ -128,7 +128,9 @@
       "voice": "en_us-lessac-medium",
       "model": "tts-1",
       "format": "mp3",
-      "speed": 1.0
+      "speed": 1.0,
+      "exaggeration": 0.5,
+      "cfg_weight": 0.5
     }
   },
   "heartbeat": {

pkg/config/config.go

@@ -217,12 +217,14 @@ type WhisperConfig struct {
 }
 
 type TTSConfig struct {
-	Enabled bool    `json:"enabled" env:"PICOCLAW_TOOLS_TTS_ENABLED"`
-	APIBase string  `json:"api_base" env:"PICOCLAW_TOOLS_TTS_API_BASE"`
-	Voice   string  `json:"voice" env:"PICOCLAW_TOOLS_TTS_VOICE"`
-	Model   string  `json:"model" env:"PICOCLAW_TOOLS_TTS_MODEL"`
-	Format  string  `json:"format" env:"PICOCLAW_TOOLS_TTS_FORMAT"`
-	Speed   float64 `json:"speed" env:"PICOCLAW_TOOLS_TTS_SPEED"`
+	Enabled      bool    `json:"enabled" env:"PICOCLAW_TOOLS_TTS_ENABLED"`
+	APIBase      string  `json:"api_base" env:"PICOCLAW_TOOLS_TTS_API_BASE"`
+	Voice        string  `json:"voice" env:"PICOCLAW_TOOLS_TTS_VOICE"`
+	Model        string  `json:"model" env:"PICOCLAW_TOOLS_TTS_MODEL"`
+	Format       string  `json:"format" env:"PICOCLAW_TOOLS_TTS_FORMAT"`
+	Speed        float64 `json:"speed" env:"PICOCLAW_TOOLS_TTS_SPEED"`
+	Exaggeration float64 `json:"exaggeration" env:"PICOCLAW_TOOLS_TTS_EXAGGERATION"` // Chatterbox: emotion expressiveness 0.0–1.0
+	CFGWeight    float64 `json:"cfg_weight" env:"PICOCLAW_TOOLS_TTS_CFG_WEIGHT"`     // Chatterbox: voice guidance weight 0.0–1.0
 }
 
 type ToolsConfig struct {
@@ -343,12 +345,14 @@ func DefaultConfig() *Config {
 				APIBase: "http://localhost:8200",
 			},
 			TTS: TTSConfig{
-				Enabled: false,
-				APIBase: "http://localhost:8100",
-				Voice:   "en_us-lessac-medium",
-				Model:   "tts-1",
-				Format:  "mp3",
-				Speed:   1.0,
+				Enabled:      false,
+				APIBase:      "http://localhost:8100",
+				Voice:        "en_us-lessac-medium",
+				Model:        "tts-1",
+				Format:       "mp3",
+				Speed:        1.0,
+				Exaggeration: 0.5,
+				CFGWeight:    0.5,
 			},
 		},
 		Heartbeat: HeartbeatConfig{

pkg/voice/kokoro.go

@@ -8,23 +8,28 @@ import (
 	"io"
 	"net/http"
 	"os"
+	"strings"
 	"time"
 
 	"github.com/sipeed/picoclaw/pkg/logger"
 )
 
 // KokoroSynthesizer uses any OpenAI-compatible /v1/audio/speech endpoint
-// (Kokoro, Piper, Chatterbox, OpenAI, etc.).
+// (Kokoro, Piper, OpenAI, etc.) and also supports Chatterbox's native
+// /synthesize endpoint for exaggeration and cfg_weight control.
 type KokoroSynthesizer struct {
-	apiBase    string
-	voice      string
-	model      string
-	format     string
-	speed      float64
-	httpClient *http.Client
+	apiBase      string
+	voice        string
+	model        string
+	format       string
+	speed        float64
+	exaggeration float64
+	cfgWeight    float64
+	httpClient   *http.Client
 }
 
-type kokoroRequest struct {
+// openaiRequest is the body for the standard /v1/audio/speech endpoint.
+type openaiRequest struct {
 	Model  string  `json:"model"`
 	Input  string  `json:"input"`
 	Voice  string  `json:"voice"`
@@ -32,13 +37,25 @@ type kokoroRequest struct {
 	Speed  float64 `json:"speed,omitempty"`
 }
 
+// chatterboxRequest is the body for Chatterbox's native /synthesize endpoint.
+// Used when model starts with "chatterbox" — gives access to emotion controls.
+type chatterboxRequest struct {
+	Text         string  `json:"text"`
+	Voice        string  `json:"voice,omitempty"`
+	Exaggeration float64 `json:"exaggeration"`
+	CFGWeight    float64 `json:"cfg_weight"`
+	Format       string  `json:"format,omitempty"`
+}
+
 // TTSProfile holds the full voice profile for the synthesizer.
 type TTSProfile struct {
-	APIBase string
-	Voice   string
-	Model   string
-	Format  string
-	Speed   float64
+	APIBase      string
+	Voice        string
+	Model        string
+	Format       string
+	Speed        float64
+	Exaggeration float64 // Chatterbox only: emotion expressiveness 0.0–1.0
+	CFGWeight    float64 // Chatterbox only: voice guidance weight 0.0–1.0
 }
 
 // NewKokoroSynthesizer creates a TTS client from a voice profile.
@@ -67,21 +84,31 @@ func NewKokoroSynthesizerFromProfile(p TTSProfile) *KokoroSynthesizer {
 	if p.Speed == 0 {
 		p.Speed = 1.0
 	}
+	if p.Exaggeration == 0 {
+		p.Exaggeration = 0.5
+	}
+	if p.CFGWeight == 0 {
+		p.CFGWeight = 0.5
+	}
 
 	logger.InfoCF("voice", "Creating TTS synthesizer", map[string]interface{}{
-		"api_base": p.APIBase,
-		"voice":    p.Voice,
-		"model":    p.Model,
-		"format":   p.Format,
-		"speed":    p.Speed,
+		"api_base":     p.APIBase,
+		"voice":        p.Voice,
+		"model":        p.Model,
+		"format":       p.Format,
+		"speed":        p.Speed,
+		"exaggeration": p.Exaggeration,
+		"cfg_weight":   p.CFGWeight,
 	})
 
 	return &KokoroSynthesizer{
-		apiBase: p.APIBase,
-		voice:   p.Voice,
-		model:   p.Model,
-		format:  p.Format,
-		speed:   p.Speed,
+		apiBase:      p.APIBase,
+		voice:        p.Voice,
+		model:        p.Model,
+		format:       p.Format,
+		speed:        p.Speed,
+		exaggeration: p.Exaggeration,
+		cfgWeight:    p.CFGWeight,
 		httpClient: &http.Client{
 			Timeout: 60 * time.Second,
 		},
@@ -90,26 +117,52 @@ func NewKokoroSynthesizerFromProfile(p TTSProfile) *KokoroSynthesizer {
 
 // Synthesize converts text to audio, writes it to a temp file, and returns the path.
 // The caller must delete the file when done.
+// isChatterbox returns true when the configured model targets the Chatterbox
+// server, which exposes a richer /synthesize endpoint alongside the standard
+// /v1/audio/speech one.
+func (s *KokoroSynthesizer) isChatterbox() bool {
+	return strings.HasPrefix(strings.ToLower(s.model), "chatterbox")
+}
+
 func (s *KokoroSynthesizer) Synthesize(ctx context.Context, text string) (string, error) {
 	logger.InfoCF("voice", "Synthesizing speech", map[string]interface{}{
-		"text_length": len(text),
-		"voice":       s.voice,
+		"text_length":  len(text),
+		"voice":        s.voice,
+		"model":        s.model,
+		"chatterbox":   s.isChatterbox(),
 	})
 
-	reqBody := kokoroRequest{
-		Model:  s.model,
-		Input:  text,
-		Voice:  s.voice,
-		Format: s.format,
-		Speed:  s.speed,
-	}
+	var (
+		bodyBytes []byte
+		url       string
+		err       error
+	)
 
-	bodyBytes, err := json.Marshal(reqBody)
+	if s.isChatterbox() {
+		// Chatterbox native endpoint — supports exaggeration and cfg_weight.
+		url = s.apiBase + "/synthesize"
+		bodyBytes, err = json.Marshal(chatterboxRequest{
+			Text:         text,
+			Voice:        s.voice,
+			Exaggeration: s.exaggeration,
+			CFGWeight:    s.cfgWeight,
+			Format:       s.format,
+		})
+	} else {
+		// Standard OpenAI-compatible endpoint.
+		url = s.apiBase + "/v1/audio/speech"
+		bodyBytes, err = json.Marshal(openaiRequest{
+			Model:  s.model,
+			Input:  text,
+			Voice:  s.voice,
+			Format: s.format,
+			Speed:  s.speed,
+		})
+	}
 	if err != nil {
 		return "", fmt.Errorf("failed to marshal TTS request: %w", err)
 	}
 
-	url := s.apiBase + "/v1/audio/speech"
 	req, err := http.NewRequestWithContext(ctx, "POST", url, bytes.NewReader(bodyBytes))
 	if err != nil {
 		return "", fmt.Errorf("failed to create TTS request: %w", err)
@@ -124,10 +177,9 @@ func (s *KokoroSynthesizer) Synthesize(ctx context.Context, text string) (string
 
 	if resp.StatusCode != http.StatusOK {
 		body, _ := io.ReadAll(resp.Body)
-		return "", fmt.Errorf("Kokoro TTS error (status %d): %s", resp.StatusCode, string(body))
+		return "", fmt.Errorf("TTS error (status %d): %s", resp.StatusCode, string(body))
 	}
 
-	// Write audio to temp file
 	tmpFile, err := os.CreateTemp("", "picoclaw-tts-*."+s.format)
 	if err != nil {
 		return "", fmt.Errorf("failed to create temp audio file: %w", err)