whisper

package
v0.0.20 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Aug 11, 2024 License: Apache-2.0 Imports: 11 Imported by: 2

Documentation

Index

Constants

View Source 
const (
	SampleRate = C.WHISPER_SAMPLE_RATE
)

Variables

View Source 
var (
	ErrTranscriptionFailed = errors.New("whisper_full failed")
)

Functions

func NewClient 

func NewClient(abspath string) *client

Create a new client with the specified root URL for downloading models

func Whisper_free added in v0.0.9 

func Whisper_free(ctx *Context)

Frees all memory allocated by the model.

func Whisper_full added in v0.0.9 

func Whisper_full(ctx *Context, params FullParams, samples []float32) error

Run the entire model: PCM -> log mel spectrogram -> encoder -> decoder -> text Not thread safe for same context Uses the specified decoding strategy to obtain the text.

func Whisper_is_multilingual added in v0.0.9 

func Whisper_is_multilingual(ctx *Context) bool

Return model capabilities

func Whisper_lang_id added in v0.0.9 

func Whisper_lang_id(str string) int

Return the language id of the specified short or full string (e.g. "de" -> 2), or -1 if not found

func Whisper_lang_max_id 

func Whisper_lang_max_id() int

Return largest language id (i.e. number of available languages - 1)

func Whisper_lang_str 

func Whisper_lang_str(id int) string

Return the short string of the specified language id (e.g. 2 -> "de"), or empty string if not found

func Whisper_lang_str_full added in v0.0.9 

func Whisper_lang_str_full(id int) string

Return the long string of the specified language name (e.g. 2 -> "german"), or empty string if not found

func Whisper_log added in v0.0.9 

func Whisper_log(level LogLevel, text string, user_data unsafe.Pointer)

Call logging output

func Whisper_log_set 

func Whisper_log_set(fn func(level LogLevel, text string))

Set logging output

Types

type AbortCallback added in v0.0.9 

type AbortCallback func() bool

If it returns true, the computation is aborted

type Client 

type Client interface {
	// Get a file from the server, writing the response to the writer
	// and returning the number of bytes copied
	Get(ctx context.Context, w io.Writer, path string) (int64, error)
}

The client interface is used to download models

type Context 

type Context C.struct_whisper_context

func Whisper_init_from_buffer_with_params added in v0.0.9 

func Whisper_init_from_buffer_with_params(data []byte, params ContextParams) *Context

Create a new context with model data and context parameters. Returns nil on error.

func Whisper_init_from_file_with_params added in v0.0.9 

func Whisper_init_from_file_with_params(path string, params ContextParams) *Context

Create a new context with path to model and context parameters. Returns nil on error.

func (*Context) DefaultLangId added in v0.0.9 

func (ctx *Context) DefaultLangId() int

Language id associated with the context's default state

func (*Context) NumSegments added in v0.0.9 

func (ctx *Context) NumSegments() int

Number of generated text segments A segment can be a few words, a sentence, or even a paragraph.

func (*Context) Segment added in v0.0.9 

func (ctx *Context) Segment(n int) *Segment

Return a segment, or nil

func (*Context) SegmentNumTokens added in v0.0.9 

func (ctx *Context) SegmentNumTokens(n int) int

Get number of tokens in the specified segment

func (*Context) SegmentSpeakerTurnNext added in v0.0.9 

func (ctx *Context) SegmentSpeakerTurnNext(n int) bool

Get whether the next segment is predicted as a speaker turn

func (*Context) SegmentT0 added in v0.0.9 

func (ctx *Context) SegmentT0(n int) int64

Get the start time of the specified segment

func (*Context) SegmentT1 added in v0.0.9 

func (ctx *Context) SegmentT1(n int) int64

Get the end time of the specified segment

func (*Context) SegmentText added in v0.0.9 

func (ctx *Context) SegmentText(n int) string

Get the text of the specified segment

func (*Context) SegmentTokenData added in v0.0.9 

func (ctx *Context) SegmentTokenData(n, i int) TokenData

Get token data for the specified token in the specified segment This contains probabilities, timestamps, etc.

func (*Context) SegmentTokenId added in v0.0.9 

func (ctx *Context) SegmentTokenId(n, i int) int32

Get the token id in the specified segment

func (*Context) SegmentTokenProb added in v0.0.9 

func (ctx *Context) SegmentTokenProb(n, i int) float32

Get the token probability of the specified token in the specified segment

func (*Context) SegmentTokenText added in v0.0.9 

func (ctx *Context) SegmentTokenText(n, i int) string

Get the token text in the specified segment

func (*Context) Token added in v0.0.9 

func (ctx *Context) Token(data TokenData) Token

Return a token from TokenData

func (*Context) Tokens added in v0.0.9 

func (ctx *Context) Tokens(n int) []Token

Return tokens for a segment

type ContextParams added in v0.0.9 

type ContextParams C.struct_whisper_context_params

func DefaultContextParams added in v0.0.9 

func DefaultContextParams() ContextParams

func (ContextParams) MarshalJSON added in v0.0.9 

func (ctx ContextParams) MarshalJSON() ([]byte, error)

func (*ContextParams) SetFlashAttn added in v0.0.9 

func (ctx *ContextParams) SetFlashAttn(v bool)

func (*ContextParams) SetGpuDevice added in v0.0.9 

func (ctx *ContextParams) SetGpuDevice(v int)

func (*ContextParams) SetTokenTimestamps added in v0.0.9 

func (ctx *ContextParams) SetTokenTimestamps(v bool)

func (*ContextParams) SetUseGpu added in v0.0.9 

func (ctx *ContextParams) SetUseGpu(v bool)

func (ContextParams) String added in v0.0.9 

func (ctx ContextParams) String() string

type FullParams added in v0.0.9 

type FullParams C.struct_whisper_full_params

func DefaultFullParams added in v0.0.9 

func DefaultFullParams(strategy SamplingStrategy) FullParams

func (*FullParams) Diarize added in v0.0.15 

func (c *FullParams) Diarize() bool

func (*FullParams) Language added in v0.0.14 

func (c *FullParams) Language() string

func (FullParams) MarshalJSON added in v0.0.9 

func (ctx FullParams) MarshalJSON() ([]byte, error)

func (*FullParams) SetAbortCallback added in v0.0.9 

func (c *FullParams) SetAbortCallback(ctx *Context, cb AbortCallback)

func (*FullParams) SetDiarize added in v0.0.15 

func (c *FullParams) SetDiarize(v bool)

func (*FullParams) SetDurationMS added in v0.0.9 

func (c *FullParams) SetDurationMS(v int)

func (*FullParams) SetLanguage added in v0.0.9 

func (c *FullParams) SetLanguage(v string)

func (*FullParams) SetMaxTextCtx added in v0.0.9 

func (c *FullParams) SetMaxTextCtx(v int)

func (*FullParams) SetNoContext added in v0.0.9 

func (c *FullParams) SetNoContext(v bool)

func (*FullParams) SetNoTimestamps added in v0.0.9 

func (c *FullParams) SetNoTimestamps(v bool)

func (*FullParams) SetNumThreads added in v0.0.9 

func (c *FullParams) SetNumThreads(v int)

func (*FullParams) SetOffsetMS added in v0.0.9 

func (c *FullParams) SetOffsetMS(v int)

func (*FullParams) SetPrintProgress added in v0.0.9 

func (c *FullParams) SetPrintProgress(v bool)

func (*FullParams) SetPrintRealtime added in v0.0.9 

func (c *FullParams) SetPrintRealtime(v bool)

func (*FullParams) SetPrintSpecial added in v0.0.9 

func (c *FullParams) SetPrintSpecial(v bool)

func (*FullParams) SetPrintTimestamps added in v0.0.9 

func (c *FullParams) SetPrintTimestamps(v bool)

func (*FullParams) SetProgressCallback added in v0.0.9 

func (c *FullParams) SetProgressCallback(ctx *Context, cb ProgressCallback)

func (*FullParams) SetSegmentCallback added in v0.0.9 

func (c *FullParams) SetSegmentCallback(ctx *Context, cb SegmentCallback)

func (*FullParams) SetSingleSegment added in v0.0.9 

func (c *FullParams) SetSingleSegment(v bool)

func (*FullParams) SetTokenTimestamps added in v0.0.9 

func (c *FullParams) SetTokenTimestamps(v bool)

func (*FullParams) SetTranslate added in v0.0.9 

func (c *FullParams) SetTranslate(v bool)

func (FullParams) String added in v0.0.9 

func (ctx FullParams) String() string

func (*FullParams) Translate added in v0.0.15 

func (c *FullParams) Translate() bool

type HTTPError added in v0.0.9 

type HTTPError struct {
	Code    int
	Message string
}

func (*HTTPError) Error added in v0.0.9 

func (e *HTTPError) Error() string

type LogLevel 

type LogLevel C.enum_ggml_log_level
const (
	LogLevelDebug LogLevel = C.GGML_LOG_LEVEL_DEBUG
	LogLevelInfo  LogLevel = C.GGML_LOG_LEVEL_INFO
	LogLevelWarn  LogLevel = C.GGML_LOG_LEVEL_WARN
	LogLevelError LogLevel = C.GGML_LOG_LEVEL_ERROR
)

func (LogLevel) String 

func (v LogLevel) String() string

type ProgressCallback added in v0.0.9 

type ProgressCallback func(progress int)

Returns the new segment number

type SamplingStrategy 

type SamplingStrategy C.enum_whisper_sampling_strategy
const (
	SAMPLING_GREEDY      SamplingStrategy = C.WHISPER_SAMPLING_GREEDY      // similar to OpenAI's GreedyDecoder
	SAMPLING_BEAM_SEARCH SamplingStrategy = C.WHISPER_SAMPLING_BEAM_SEARCH // similar to OpenAI's BeamSearchDecoder
)

func (SamplingStrategy) MarshalJSON 

func (v SamplingStrategy) MarshalJSON() ([]byte, error)

func (SamplingStrategy) String 

func (v SamplingStrategy) String() string

type Segment added in v0.0.9 

type Segment struct {
	Id          int32         `json:"id"`
	Text        string        `json:"text,omitempty"`
	T0          time.Duration `json:"t0,omitempty"`
	T1          time.Duration `json:"t1,omitempty"`
	SpeakerTurn bool          `json:"speaker_turn,omitempty"`
	Tokens      []Token       `json:"tokens,omitempty"`
}

func (Segment) String added in v0.0.9 

func (s Segment) String() string

type SegmentCallback added in v0.0.9 

type SegmentCallback func(segment int)

Returns the new segment number

type Token 

type Token struct {
	Id   int32         `json:"id"`
	Text string        `json:"text,omitempty"`
	P    float32       `json:"p,omitempty"`
	T0   time.Duration `json:"t0,omitempty"`
	T1   time.Duration `json:"t1,omitempty"`
	Type TokenType     `json:"type,omitempty"`
}

func (Token) String added in v0.0.9 

func (t Token) String() string

type TokenData 

type TokenData C.struct_whisper_token_data

type TokenType added in v0.0.9 

type TokenType int
const (
	EOT  TokenType // end of text
	SOT            // start of transcript
	SOLM           // start of language model
	PREV           // start of previous segment
	NOSP           // no speaker
	NOT            // no timestamps
	BEG            // begin
	LANG           // language (TODO)
)

func (TokenType) MarshalJSON added in v0.0.9 

func (t TokenType) MarshalJSON() ([]byte, error)

func (TokenType) String added in v0.0.9 

func (t TokenType) String() string

type Writer added in v0.0.9 

type Writer interface {
	io.Writer

	// Returns the header of the response. If the return value is
	// not nil, then the Get method will end before the response
	// data is written
	Header(http.Header) error
}

If the writer contains a Header method, it can be used to set the content type and length of the response, to measure progress

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.