deepgram

type Config struct {
	// APIKey is the Deepgram API key. Required.
	APIKey string `validate:"required"`
	// ListenURL overrides the live-transcription WebSocket endpoint; empty uses
	// Deepgram's hosted endpoint.
	ListenURL string
	// Model is the Deepgram model; empty uses "nova-3".
	Model string
	// Language is the transcription language; empty uses English (US).
	Language language.Language
	// SampleRate is the input audio sample rate; 0 uses the transport's rate.
	SampleRate int
	// Encoding is the audio encoding; empty uses "linear16".
	Encoding string
	// Channels is the channel count; 0 uses 1.
	Channels int

	// InterimResults emits partial transcripts; nil defaults to true.
	InterimResults *bool
	// SmartFormat applies Deepgram smart formatting; nil defaults to true
	// (Deepgram's own default is false).
	SmartFormat *bool
	// Punctuate adds punctuation; nil defaults to true.
	Punctuate *bool
	// VADEvents requests speech-start/utterance events; nil defaults to true.
	VADEvents *bool
	// Endpointing is the silence in ms before a result is finalized; nil
	// defaults to 300 (Deepgram's own default is 10).
	Endpointing *int
	// UtteranceEndMs is the silence in ms before an utterance-end event; nil
	// defaults to 1000 (Deepgram leaves it unset).
	UtteranceEndMs *int

	// Numerals converts spoken numbers to numerals.
	Numerals *bool
	// ProfanityFilter filters profanity from transcripts.
	ProfanityFilter *bool
	// Diarize labels speakers in the transcript.
	Diarize *bool
	// DetectEntities enables named-entity detection.
	DetectEntities *bool
	// Dictation converts spoken commands (e.g. "comma") to punctuation.
	Dictation *bool
	// Multichannel transcribes each channel independently.
	Multichannel *bool
	// MipOptOut opts out of Deepgram's model-improvement program.
	MipOptOut *bool
	// Version pins the model version (e.g. "latest").
	Version string
	// Keywords boosts recognition of the given terms (legacy models); each entry
	// may carry an intensifier, e.g. "snuffleupagus:5".
	Keywords []string
	// Keyterm boosts recognition of the given terms (nova-3).
	Keyterm []string
	// Redact removes the given categories of sensitive data.
	Redact []string
	// Replace applies find/replace rules to the transcript.
	Replace []string
	// Search flags occurrences of the given terms.
	Search []string
	// Tag attaches billing tags to the request.
	Tag []string
	// ExtraQuery sets arbitrary additional Deepgram query parameters not modeled
	// above; values override any param of the same name set from other fields.
	ExtraQuery map[string]string
}

type TTSConfig struct {
	// APIKey is the Deepgram API key. Required.
	APIKey string `validate:"required"`
	// Model is the Aura voice model; empty uses a default.
	Model string
	// SampleRate is the PCM rate requested from Aura and emitted downstream;
	// 0 uses 24 kHz.
	SampleRate int
	// Encoding is the audio encoding; empty uses "linear16".
	Encoding string
}