synthesizer

Documentation

Index

• func GetTranscriberAccent(w *ProcessedWord) int
• type AccentVariant
• type AudioDurations
• type Clitic
• type CliticAccentEnum
• type MainWorker
  • func (mw *MainWorker) Add(pr Processor)
  • func (mw *MainWorker) AddSSML(pr Processor)
  • func (mw *MainWorker) GetProcessorsInfo() string
  • func (mw *MainWorker) GetSSMLProcessorsInfo() string
  • func (mw *MainWorker) Work(input *api.TTSRequestConfig) (*api.Result, error)
• type PartProcessor
• type PartRunner
  • func NewPartRunner(parallelWorker int) *PartRunner
  • func (p *PartRunner) Add(pr PartProcessor)
  • func (p *PartRunner) Process(data *TTSData) error
• type ProcessedWord
• type Processor
• type SSMLTypeEnum
• type SynthesizedPos
• type TTSConfig
• type TTSData
• type TTSDataPart
• type TTSTextPart
• type TaggedWord
  • func (tw TaggedWord) IsWord() bool

Functions

func GetTranscriberAccent

func GetTranscriberAccent(w *ProcessedWord) int

GetTranscriberAccent return accent from ProcessedWord

Types

type AccentVariant

type AccentVariant struct {
	Accent   int     `json:"accent"`
	Accented string  `json:"accented"`
	Ml       string  `json:"ml"`
	Syll     string  `json:"syll"`
	Usage    float64 `json:"usage"`
}

AccentVariant - accenters's result

type AudioDurations

type AudioDurations struct {
	// in tts steps
	Shift    int
	Duration time.Duration
}

type Clitic

type Clitic struct {
	Type   CliticAccentEnum
	Accent int
}

Clitic structure

type CliticAccentEnum

type CliticAccentEnum int

CliticAccentEnum contains types of possible clitics

const (
	//CliticsUnused - clitics does not apply for the word
	CliticsUnused CliticAccentEnum = iota

	//CliticsNone - not a clitic
	CliticsNone

	//CliticsCustom - custom clitic type
	CliticsCustom
)

type MainWorker

type MainWorker struct {
	AllowCustomCode bool
	// contains filtered or unexported fields
}

MainWorker does synthesis work

func (*MainWorker) Add

func (mw *MainWorker) Add(pr Processor)

Add adds a processor to the end

func (*MainWorker) AddSSML

func (mw *MainWorker) AddSSML(pr Processor)

AddSSML adds a SSML processor to the end

func (*MainWorker) GetProcessorsInfo

func (mw *MainWorker) GetProcessorsInfo() string

GetProcessorsInfo return info about processors for testing

func (*MainWorker) GetSSMLProcessorsInfo

func (mw *MainWorker) GetSSMLProcessorsInfo() string

GetSSMLProcessorsInfo return info about processors for testing

func (*MainWorker) Work

func (mw *MainWorker) Work(input *api.TTSRequestConfig) (*api.Result, error)

Work is main method

type PartProcessor

type PartProcessor interface {
	Process(*TTSDataPart) error
}

PartProcessor interface

type PartRunner

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

PartRunner runs parts of the job

func NewPartRunner

func NewPartRunner(parallelWorker int) *PartRunner

NewPartRunner creates parallel runner

func (*PartRunner) Add

func (p *PartRunner) Add(pr PartProcessor)

Add adds a processor to the end

func (*PartRunner) Process

func (p *PartRunner) Process(data *TTSData) error

Process is main method

type ProcessedWord

type ProcessedWord struct {
	Tagged            TaggedWord
	UserTranscription string
	UserSyllables     string
	TranscriptionWord string
	AccentVariant     *AccentVariant
	UserAccent        int
	Clitic            Clitic
	Transcription     string
	Obscene           bool
	TextPart          *TTSTextPart
	SynthesizedPos    *SynthesizedPos
}

ProcessedWord keeps one word info

type Processor

type Processor interface {
	Process(*TTSData) error
}

Processor interface

type SSMLTypeEnum

type SSMLTypeEnum int

SSMLTypeEnum indicates part type: text, pause

const (
	// SSMLNone - not ssml part
	SSMLNone SSMLTypeEnum = iota
	// SSMLMain - main part
	SSMLMain
	// SSMLText - text part for synthesis
	SSMLText
	// SSMLPause - <p>, <break> part for synthesis
	SSMLPause
)

type SynthesizedPos

type SynthesizedPos struct {
	// in tts steps
	From, To int
}

type TTSConfig

type TTSConfig struct {
	JustAM bool
	Input  *api.TTSRequestConfig

	Type  SSMLTypeEnum
	Voice string
	Speed float32

	PauseDuration time.Duration
}

TTSConfig some TTS configuration

type TTSData

type TTSData struct {
	Input        *api.TTSRequestConfig
	RequestID    string
	Cfg          TTSConfig
	OriginalText string
	PreviousText string // text of previous request loaded by requestID

	CleanedText []string // corresponds to OriginalTextParts

	NormalizedText  []string // text after normalization, array corresponds to OriginalTextParts
	Text            []string // text after cleaning and URL replacement
	TextWithNumbers []string

	AudioSuffix string // add audio suffix if var is set

	Words []*ProcessedWord
	Parts []*TTSDataPart

	Audio           string
	AudioMP3        string
	AudioLenSeconds float64
	SampleRate      uint32

	OriginalTextParts []*TTSTextPart
	SSMLParts         []*TTSData
}

TTSData working data for one request

type TTSDataPart

type TTSDataPart struct {
	Text               string
	Cfg                *TTSConfig
	First              bool
	Words              []*ProcessedWord
	Spectogram         string
	Audio              string
	TranscribedText    string
	TranscribedSymbols []string
	// from AM response
	Durations      []int
	DefaultSilence int
	Step           int
	AudioDurations *AudioDurations
}

TTSDataPart partial tts data

type TTSTextPart

type TTSTextPart struct {
	Accented, Text, Syllables, UserOEPal string
}

TTSTextPart part of the text

type TaggedWord

type TaggedWord struct {
	Separator   string
	SentenceEnd bool
	Space       bool
	Word        string
	Mi          string
	Lemma       string
}

TaggedWord - tagger's result

func (TaggedWord) IsWord

func (tw TaggedWord) IsWord() bool

IsWord returns true if object indicates word