stt

Documentation

Overview

Package stt provides OmniVoice-based speech-to-text for marp2video.

Package stt provides OmniVoice-based speech-to-text for marp2video. This file re-exports the omnivoice subtitle package for convenience.

Index

• Constants
• func GenerateSRTFromResult(result *TranscriptionResult, opts SubtitleOptions) string
• func GenerateVTTFromResult(result *TranscriptionResult, opts SubtitleOptions) string
• func SaveSRTFromResult(result *TranscriptionResult, filePath string, opts SubtitleOptions) error
• func SaveVTTFromResult(result *TranscriptionResult, filePath string, opts SubtitleOptions) error
• type Factory
  • func NewFactory(config ProviderConfig) *Factory
  • func (f *Factory) Available() []string
  • func (f *Factory) Get(name string) (*Provider, error)
  • func (f *Factory) SetFallback(name string)
• type Provider
  • func New(provider omnivoice.STTProvider) *Provider
  • func (p *Provider) Name() string
  • func (p *Provider) Transcribe(ctx context.Context, audio []byte, config TranscriptionConfig) (*TranscriptionResult, error)
  • func (p *Provider) TranscribeFile(ctx context.Context, filePath string, config TranscriptionConfig) (*TranscriptionResult, error)
  • func (p *Provider) TranscribeURL(ctx context.Context, url string, config TranscriptionConfig) (*TranscriptionResult, error)
  • func (p *Provider) UnderlyingProvider() omnivoice.STTProvider
• type ProviderConfig
• type Segment
• type SubtitleFormat
• type SubtitleOptions
  • func DefaultSubtitleOptions() SubtitleOptions
• type TranscriptionConfig
• type TranscriptionResult
• type Word

Constants

const (
	FormatSRT = omnivoice.SubtitleFormatSRT
	FormatVTT = omnivoice.SubtitleFormatVTT
)

Format constants.

Functions

func GenerateSRTFromResult

func GenerateSRTFromResult(result *TranscriptionResult, opts SubtitleOptions) string

GenerateSRTFromResult generates SRT from a local TranscriptionResult. For direct omnivoice results, use omnivoice.GenerateSRT instead.

func GenerateVTTFromResult

func GenerateVTTFromResult(result *TranscriptionResult, opts SubtitleOptions) string

GenerateVTTFromResult generates VTT from a local TranscriptionResult. For direct omnivoice results, use omnivoice.GenerateVTT instead.

func SaveSRTFromResult

func SaveSRTFromResult(result *TranscriptionResult, filePath string, opts SubtitleOptions) error

SaveSRTFromResult saves SRT from a local TranscriptionResult.

func SaveVTTFromResult

func SaveVTTFromResult(result *TranscriptionResult, filePath string, opts SubtitleOptions) error

SaveVTTFromResult saves VTT from a local TranscriptionResult.

Types

type Factory

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

Factory creates STT providers based on configuration.

func NewFactory

func NewFactory(config ProviderConfig) *Factory

NewFactory creates a new STT provider factory.

func (*Factory) Available

func (f *Factory) Available() []string

Available returns a list of available provider names based on configured API keys.

func (*Factory) Get

func (f *Factory) Get(name string) (*Provider, error)

Get returns a provider by name, creating it if necessary. If name is empty, returns the fallback provider.

func (*Factory) SetFallback

func (f *Factory) SetFallback(name string)

SetFallback sets the default provider name.

type Provider

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

Provider wraps an OmniVoice STT provider for use with marp2video.

func New

func New(provider omnivoice.STTProvider) *Provider

New creates a new OmniVoice STT provider wrapper.

func (*Provider) Name

func (p *Provider) Name() string

Name returns the provider name.

func (*Provider) Transcribe

func (p *Provider) Transcribe(ctx context.Context, audio []byte, config TranscriptionConfig) (*TranscriptionResult, error)

Transcribe transcribes audio bytes and returns the result.

func (*Provider) TranscribeFile

func (p *Provider) TranscribeFile(ctx context.Context, filePath string, config TranscriptionConfig) (*TranscriptionResult, error)

TranscribeFile transcribes an audio file and returns the result.

func (*Provider) TranscribeURL

func (p *Provider) TranscribeURL(ctx context.Context, url string, config TranscriptionConfig) (*TranscriptionResult, error)

TranscribeURL transcribes audio from a URL and returns the result.

func (*Provider) UnderlyingProvider

func (p *Provider) UnderlyingProvider() omnivoice.STTProvider

UnderlyingProvider returns the wrapped OmniVoice provider for advanced operations.

type ProviderConfig

type ProviderConfig struct {
	// ElevenLabsAPIKey is the API key for ElevenLabs.
	ElevenLabsAPIKey string

	// DeepgramAPIKey is the API key for Deepgram.
	DeepgramAPIKey string
}

ProviderConfig holds configuration for creating STT providers.

type Segment

type Segment struct {
	// Text is the transcribed text for this segment.
	Text string

	// StartTime is when the segment starts.
	StartTime time.Duration

	// EndTime is when the segment ends.
	EndTime time.Duration

	// Confidence is the average confidence for this segment.
	Confidence float64

	// Speaker is the speaker identifier (if diarization enabled).
	Speaker string

	// Words contains word-level details.
	Words []Word
}

Segment represents a segment of transcription.

type SubtitleFormat

type SubtitleFormat = omnivoice.SubtitleFormat

SubtitleFormat is an alias for omnivoice.SubtitleFormat.

type SubtitleOptions

type SubtitleOptions = omnivoice.SubtitleOptions

SubtitleOptions is an alias for omnivoice.SubtitleOptions.

func DefaultSubtitleOptions

func DefaultSubtitleOptions() SubtitleOptions

DefaultSubtitleOptions returns sensible defaults for subtitle generation.

type TranscriptionConfig

type TranscriptionConfig struct {
	// Language is the BCP-47 language code (e.g., "en-US").
	Language string

	// Model is the provider-specific model identifier.
	Model string

	// EnablePunctuation adds punctuation to transcription.
	EnablePunctuation bool

	// EnableWordTimestamps includes word-level timestamps.
	EnableWordTimestamps bool

	// EnableSpeakerDiarization identifies different speakers.
	EnableSpeakerDiarization bool

	// MaxSpeakers is the maximum number of speakers to detect.
	MaxSpeakers int
}

TranscriptionConfig configures a transcription request.

type TranscriptionResult

type TranscriptionResult struct {
	// Text is the full transcription text.
	Text string

	// Segments contains segment-level details with timing.
	Segments []Segment

	// Language is the detected language.
	Language string

	// Duration is the audio duration.
	Duration time.Duration
}

TranscriptionResult contains the result of a transcription.

type Word

type Word struct {
	// Text is the transcribed word.
	Text string

	// StartTime is when the word starts.
	StartTime time.Duration

	// EndTime is when the word ends.
	EndTime time.Duration

	// Confidence is the recognition confidence (0.0 to 1.0).
	Confidence float64

	// Speaker is the speaker identifier.
	Speaker string
}

Word represents a single transcribed word with timing.