Support Gemini via Cloud TTS API with style and language params, add Google gcloud auth support for same, and improve setup flow

Author: edumansky

CLAUDE.md

@@ -18,7 +18,8 @@ go vet ./...        # Static analysis
 - `internal/config/` — Runtime state (debug, dry-run flags)
 - `internal/output/` — JSON/table/CSV/raw output formatting
 - `internal/keyring/` — API key storage in OS keyring
-- `internal/provider/` — TTSProvider and STTProvider interfaces, registry
+- `internal/provider/` — TTSProvider and STTProvider interfaces, registry, AuthConfig
+- `internal/gcpauth/` — GCP Application Default Credentials resolution and token exchange (no SDK)
 - `internal/tts/` — TTS provider implementations (Google, ElevenLabs, OpenAI)
 - `internal/stt/` — STT provider implementations (Google, AssemblyAI, OpenAI)
 - `internal/audio/` — WAV header generation, PCM/mulaw/alaw conversion
@@ -27,14 +28,15 @@ go vet ./...        # Static analysis
 ## Key Conventions
 
 - REST over SDK: Use net/http directly, no Google Cloud SDK
-- API key resolution: flag → env var → OS keyring
+- Auth resolution (Google): ADC/OAuth2 → env var → OS keyring; (others): env var → OS keyring
 - Default audio: 8kHz mu-law WAV (IVR standard)
 - Provider registration: init() functions in tts/ and stt/ packages register with provider.Registry
 - Output formatting: Use output.PrintObject() for structured data, fmt.Fprintf(os.Stderr, ...) for status messages
 - Config: Non-secret config in ~/.prompt-tools/config.json, secrets in OS keyring under "prompt-tools-cli" service name
 
 ## Environment Variables
 
+- `GOOGLE_APPLICATION_CREDENTIALS` — Path to GCP service account JSON key file (ADC)
 - `GOOGLE_API_KEY` — Google Cloud API key
 - `ELEVENLABS_API_KEY` — ElevenLabs API key
 - `ASSEMBLYAI_API_KEY` — AssemblyAI API key

cmd/batch_transcribe.go

@@ -84,13 +84,13 @@ var batchTranscribeCmd = &cobra.Command{
 			return nil
 		}
 
-		// Resolve API key
-		apiKey, err := resolveAPIKey(providerName)
+		// Resolve auth
+		auth, err := resolveAuth(providerName)
 		if err != nil {
 			return err
 		}
 
-		sttProvider, err := provider.NewSTT(providerName, apiKey)
+		sttProvider, err := provider.NewSTT(providerName, auth)
 		if err != nil {
 			return err
 		}

cmd/bulk.go

@@ -24,14 +24,17 @@ var bulkGenerateCmd = &cobra.Command{
 
 Input format (row 1 = headers, skipped automatically):
 
-  Filename        | Voice             | Text                  | SSML | Sample Rate | Encoding | Notes
-  welcome.wav     | en-US-Chirp3-HD-Achernar   | Welcome to support.   | no   |             |          | Main greeting
-  #holiday.wav    | en-US-Chirp3-HD-Achernar   | Closed for holiday.   | no   |             |          | Skipped
-  es/welcome.wav  | es-MX-Chirp3-HD-A | Bienvenido.           | no   |             |          | Subdirectory
+  Filename        | Voice             | Text                  | SSML | Sample Rate | Encoding | Notes         | Style              | Language
+  welcome.wav     | en-US-Chirp3-HD-…| Welcome to support.   | no   |             |          | Main greeting |                    |
+  #holiday.wav    | en-US-Chirp3-HD-…| Closed for holiday.   | no   |             |          | Skipped       |                    |
+  es/welcome.wav  | es-MX-Chirp3-HD-A| Bienvenido.           | no   |             |          | Subdirectory  |                    |
+  gemini.wav      | Achernar          | Welcome to support.  | no   |             |          | Gemini+style  | speak warmly       | en-US
 
   - Rows starting with # are skipped.
-  - Voice, Sample Rate, and Encoding are optional (defaults from config).
+  - Voice, Sample Rate, Encoding, Style, and Language are optional (defaults from config).
   - SSML column: yes/no — whether Text contains SSML markup.
+  - Style column: voice steering prompt (Gemini voices with GCP OAuth2 auth only).
+  - Language column: language code for Gemini voices (default: en-US).
   - Filename supports subdirectories (e.g., en-US/welcome.wav) — folders are created automatically.
 
 Examples:
@@ -124,13 +127,13 @@ Examples:
 			return nil
 		}
 
-		// Resolve API key and create provider
-		apiKey, err := resolveAPIKey(providerName)
+		// Resolve auth and create provider
+		auth, err := resolveAuth(providerName)
 		if err != nil {
 			return err
 		}
 
-		tts, err := provider.NewTTS(providerName, apiKey)
+		tts, err := provider.NewTTS(providerName, auth)
 		if err != nil {
 			return err
 		}

cmd/config_cmd.go

@@ -6,6 +6,7 @@ import (
 	"strings"
 
 	"github.com/Cloverhound/prompt-tools-cli/internal/appconfig"
+	"github.com/Cloverhound/prompt-tools-cli/internal/gcpauth"
 	"github.com/Cloverhound/prompt-tools-cli/internal/keyring"
 	"github.com/Cloverhound/prompt-tools-cli/internal/output"
 	"github.com/spf13/cobra"
@@ -57,6 +58,13 @@ var configShowCmd = &cobra.Command{
 			}
 		}
 
+		// Show GCP ADC status
+		if gcpauth.HasCredentials() {
+			display.APIKeys["google_adc"] = "configured"
+		} else {
+			display.APIKeys["google_adc"] = "not set"
+		}
+
 		return output.PrintObject(display)
 	},
 }
@@ -230,6 +238,26 @@ var configSetAPIKeyCmd = &cobra.Command{
 	},
 }
 
+var configSetGCPProjectCmd = &cobra.Command{
+	Use:   "set-gcp-project <project-id>",
+	Short: "Set GCP project for Google API quota/billing",
+	Args:  cobra.ExactArgs(1),
+	RunE: func(cmd *cobra.Command, args []string) error {
+		cfg, err := appconfig.Load()
+		if err != nil {
+			return err
+		}
+		gc := cfg.Providers["google"]
+		gc.ProjectID = args[0]
+		cfg.Providers["google"] = gc
+		if err := cfg.Save(); err != nil {
+			return err
+		}
+		fmt.Printf("GCP project set to %s\n", args[0])
+		return nil
+	},
+}
+
 var configClearAPIKeyCmd = &cobra.Command{
 	Use:   "clear-api-key <google|elevenlabs|assemblyai|openai>",
 	Short: "Remove API key from keyring",
@@ -266,5 +294,6 @@ func init() {
 	configCmd.AddCommand(configSetEncodingCmd)
 	configCmd.AddCommand(configSetAPIKeyCmd)
 	configCmd.AddCommand(configClearAPIKeyCmd)
+	configCmd.AddCommand(configSetGCPProjectCmd)
 	rootCmd.AddCommand(configCmd)
 }