realtime

type STTConnection struct {
	// contains filtered or unexported fields
}

type STTOptions struct {
	// ModelID is the transcription model to use.
	ModelID string

	// AudioFormat specifies the audio encoding format.
	AudioFormat string

	// LanguageCode is the expected language.
	LanguageCode string

	// IncludeTimestamps enables word-level timing information.
	IncludeTimestamps bool

	// IncludeLanguageDetection includes detected language in responses.
	IncludeLanguageDetection bool

	// CommitStrategy determines how transcripts are committed.
	CommitStrategy string

	// VAD settings
	VADSilenceThresholdSecs float64
	VADThreshold            float64
	MinSpeechDurationMs     int
	MinSilenceDurationMs    int
}

type STTTranscript struct {
	Text         string    `json:"text"`
	IsFinal      bool      `json:"is_final"`
	Words        []STTWord `json:"words,omitempty"`
	LanguageCode string    `json:"language_code,omitempty"`
}

type STTWord struct {
	Text      string  `json:"text"`
	Start     float64 `json:"start"`
	End       float64 `json:"end"`
	Type      string  `json:"type,omitempty"`
	SpeakerID string  `json:"speaker_id,omitempty"`
}

type Service struct {
	// contains filtered or unexported fields
}

type TTSAlignment struct {
	Characters     []string  `json:"characters"`
	CharacterStart []float64 `json:"character_start_times_seconds"`
	CharacterEnd   []float64 `json:"character_end_times_seconds"`
}

type TTSConnection struct {
	// contains filtered or unexported fields
}

type TTSOptions struct {
	// ModelID is the model to use. Defaults to "eleven_turbo_v2_5" for low latency.
	ModelID string

	// OutputFormat specifies the audio output format.
	// Recommended for real-time: "pcm_16000", "pcm_22050", "pcm_24000", "pcm_44100"
	OutputFormat string

	// VoiceSettings configures the voice parameters.
	VoiceSettings *VoiceSettings

	// OptimizeStreamingLatency reduces latency at the cost of quality (0-4).
	OptimizeStreamingLatency int

	// EnableSSMLParsing enables SSML parsing for the input text.
	EnableSSMLParsing bool

	// LanguageCode is the ISO language code (e.g., "en", "es").
	LanguageCode string

	// ChunkLengthSchedule controls text chunking for audio generation.
	ChunkLengthSchedule []int

	// InactivityTimeout is the context timeout in seconds (default 20).
	InactivityTimeout int

	// PronunciationDictionaryIDs is a list of pronunciation dictionary IDs to use.
	PronunciationDictionaryIDs []string
}

type VoiceSettings struct {
	Stability       float64
	SimilarityBoost float64
	Style           float64
	Speed           float64
	UseSpeakerBoost bool
}