hal

package module
v0.0.0-...-a435009 Latest Latest
Warning

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

 Go to latest Published: Apr 11, 2023 License: Apache-2.0 Imports: 16 Imported by: 0

README

HAL

HAL is a voice-enabled version of chatGPT that can run on command-line terminals, and multilingual support.

Note: the Azure Speech SDK for go only support linux, so this project only run on linux environment. The Speech SDK does not support OpenSSL 3.0, which is the default in Ubuntu 22.04. On Ubuntu 22.04 only, install the latest libssl1.1 either as a binary package, or by compiling it from sources. Read more about this.

Prerequisites

✅ Get an OpenAI API key
✅ Get an Azure Speech key

Quick Start

1. Install

require: go >= 1.18

git clone https://github.com/neotse/hal.git
cd hal
chmod +x install.sh
./install.sh
2. Configure

just run it, and following the guide to configure.

hal

If you want to reconfigure, run below command:

hal --init

Or, you can see the help:

hal --help
3. Use
Activate and Deactivate

Try to say Harold to Activate, and stopword (configured above) to Deactivate (or automatically deactivated if there is no speech for a while).

Talk

just talk to HAL in your language (configured above)

if you want custom keyword to activate

First, you need generate a keyword model file from Azure, and download the model file. Then, use hal keyword command to configure it.

Usage of keyword:
  -keyword string
        set the keyword for activate (case insensitive), path and lang must be set at same time.
  -lang string
        set the language of keyword. (see https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/language-support?tabs=stt)
  -path string
        set the path of model file of keyword.
  -show
        show the current config of keyword for activate.

Session

Most of the time, conversations have context. Therefore, retaining some context can improve the quality of chatGPT's responses. In addition, OpenAI also provides system type messages to reinforce chatGPT's attention to improve the quality of responses. Therefore, here, sessions are used to retain this information. You can manage sessions, including listing sessions, selecting sessions, editing sessions, and creating sessions. Session management can be done in two ways:

Pre-run configuration

use hal session command to configure sessions.

Usage of session:
  -config
        config the 'session' for talk. 
  -create
        create the 'session' for talk. 
  -delete
        delete the 'session' for talk. 
  -list
        list current chatgpt sessions.
  -select
        select the 'session' for start to talk. If not set, it will select the session recently used.
Configuration by voice

When running, you can say the keyword (or similar in meaning) to invoke corresponding the action.

action keyword for example (other languages are also supported)
list list sessions please list the sessions
select select session I want to select a session
delete delete session delete a session
create create session help me create a session
config configure session configure the session

Documentation

Index

Constants

View Source 
const AUDIO_FORMT = common.Audio16Khz32KBitRateMonoMp3

Variables

View Source 
var (
	ErrRepeatConfig = errors.New("found a same config")
	ErrNoSuchHook   = errors.New("hook not exists")
)
View Source 
var (
	ErrSpeechRecognitionTimeout                    = errors.New("speech recognition get result timeout")
	MaxSpeechRecognitionDelay                      = time.Duration(30)
	SegmentationSilenceTimeoutMs                   = "1500" //1.5s
	SpeechServiceConnectionInitialSilenceTimeoutMs = "5000" //5s
)
View Source 
var (
	ErrSpeechSynthesisTimeout = errors.New("speech synthesis get result timeout")
	MaxSpeechSynthesisDelay   = time.Duration(10)
)
View Source 
var AUTO_DETECTED_LANGUAGE = map[string]string{

	"English": "en-US",
	"French":  "fr-FR",
	"Chinese": "zh-CN",
	"Spanish": "es-ES",
}

because azure stt go speech sdk only support 4 languages in auto detected by languages (not support full languages auto detected)

View Source 
var CHATGPTS = newChatGPTs()
View Source 
var HOOKS = newHooks()
View Source 
var PARAMS = Params{MaxHistory: 4, Language: "en-US", Voice: "en-US-ElizabethNeural"}

Functions

func CheckOpenaiKey 

func CheckOpenaiKey(key string) error

func ConfigSession 

func ConfigSession()

func CreateASession 

func CreateASession()

func DefaultLogger 

func DefaultLogger(level Level) *defaultLog

func DeleteSession 

func DeleteSession()

func FormatVoices 

func FormatVoices(voices []*Voice) string

func GetAutoDetectedLanguages 

func GetAutoDetectedLanguages() []string

func GetAutoDetectedLanguagesName 

func GetAutoDetectedLanguagesName() []string

func Initialize 

func Initialize(chatGPTModel, language, voice, stopWord string)

func ListSessions 

func ListSessions()

func SelectSession 

func SelectSession()

func SetLevel 

func SetLevel(level Level)

func Showkeyword 

func Showkeyword()

Types

type ChatGPT 

type ChatGPT struct {
	System     *openai.ChatCompletionMessage   `json:"system"`
	History    []*openai.ChatCompletionMessage `json:"history"`
	MaxHistory int                             `json:"maxHistory"`
	Model      string                          `json:"model"`
	Key        string                          `json:"key"`
	IsDefault  bool                            `json:"default"`
	// contains filtered or unexported fields
}

func NewChatGPT 

func NewChatGPT(key string, model string) *ChatGPT

func (*ChatGPT) DisableHistory 

func (c *ChatGPT) DisableHistory()

func (*ChatGPT) Prompt 

func (c *ChatGPT) Prompt(text string) (string, int, error)

func (*ChatGPT) PromptStream 

func (c *ChatGPT) PromptStream(text string) (*StreamResult, error)

func (*ChatGPT) SetMaxHistory 

func (c *ChatGPT) SetMaxHistory(n int)

func (*ChatGPT) SetModel 

func (c *ChatGPT) SetModel(model string)

func (*ChatGPT) SetRole 

func (c *ChatGPT) SetRole(text string)

func (*ChatGPT) String 

func (c *ChatGPT) String() string

type ChatGPTs 

type ChatGPTs struct {
	Clients map[string]*ChatGPT `json:"clients"`
}

func (ChatGPTs) DelSession 

func (c ChatGPTs) DelSession(sessionName string)

func (ChatGPTs) GetDefaultGPT 

func (c ChatGPTs) GetDefaultGPT() (string, *ChatGPT)

func (*ChatGPTs) LoadChatGPTs 

func (c *ChatGPTs) LoadChatGPTs(file string) error

func (ChatGPTs) NewDefaultSession 

func (c ChatGPTs) NewDefaultSession(key string, model string) *ChatGPT

func (ChatGPTs) NewSessionWithName 

func (c ChatGPTs) NewSessionWithName(sessionName string, key string, model string) *ChatGPT

func (ChatGPTs) RenameSession 

func (c ChatGPTs) RenameSession(oldName string, newName string) *ChatGPT

func (ChatGPTs) SaveChatGPTs 

func (c ChatGPTs) SaveChatGPTs(file string) error

func (ChatGPTs) Sessions 

func (c ChatGPTs) Sessions() []string

func (ChatGPTs) SessionsWithout 

func (c ChatGPTs) SessionsWithout(name string) []string

func (ChatGPTs) SetDefaultGPT 

func (c ChatGPTs) SetDefaultGPT(sessionName string)

type Hook 

type Hook interface {
	New() Hook
	Check(stt string) bool
	SetKeyword(text string)
	Name() string
	Exec() error
}

type HookConfig 

type HookConfig struct {
	Keyword  string `json:"keyword"`
	HookName string `json:"hook"`
	Enable   bool   `json:"enable"`
	// contains filtered or unexported fields
}

type Hooks 

type Hooks struct {
	Configs map[string]*HookConfig `json:"hookConfigs"`
	// contains filtered or unexported fields
}

func (*Hooks) Add 

func (h *Hooks) Add(keyword string, name string) error

func (*Hooks) Delete 

func (h *Hooks) Delete(id string) bool

func (*Hooks) Disable 

func (h *Hooks) Disable(id string)

func (*Hooks) Enable 

func (h *Hooks) Enable(id string)

func (*Hooks) Exec 

func (h *Hooks) Exec(hookName string) error

func (*Hooks) Get 

func (h *Hooks) Get(id string) *HookConfig

func (*Hooks) IsExist 

func (h *Hooks) IsExist(hookName string) bool

func (*Hooks) LoadHooks 

func (h *Hooks) LoadHooks(file string) error

func (*Hooks) SaveHooks 

func (h *Hooks) SaveHooks(file string) error

type KeywordRecognitionStandalone 

type KeywordRecognitionStandalone struct {
	SpeechRecognitionStream

	KeyWord string
	// contains filtered or unexported fields
}

func NewKeywordRecognitionStandalone 

func NewKeywordRecognitionStandalone(key, region string, languages []string, model, keyWord string) (*KeywordRecognitionStandalone, error)

func (*KeywordRecognitionStandalone) Close 

func (s *KeywordRecognitionStandalone) Close() error

func (*KeywordRecognitionStandalone) Result 

func (s *KeywordRecognitionStandalone) Result() (string, error)

func (*KeywordRecognitionStandalone) SpeechToText 

func (s *KeywordRecognitionStandalone) SpeechToText(data []byte) error

func (*KeywordRecognitionStandalone) Start 

func (s *KeywordRecognitionStandalone) Start() error

type Level 

type Level int
const (
	DEBUG Level = iota
	INFO
	WARNING
	ERROR
	SLICENCE
)

type Logger 

type Logger interface {
	Errorf(string, ...interface{})
	Warningf(string, ...interface{})
	Infof(string, ...interface{})
	Debugf(string, ...interface{})
}

type Params 

type Params struct {
	Initialized  bool
	OpenaiKey    string
	ChatgptModel string
	MaxHistory   int

	SpeechKey       string
	SpeechRegion    string
	Language        string // BCP-47 code
	Voice           string
	StopWord        string
	Keyword         string
	KeywordModel    string
	KeywordLanguage string
}

func (*Params) LoadParams 

func (p *Params) LoadParams(file string) error

func (Params) SaveParams 

func (p Params) SaveParams(file string) error

func (Params) String 

func (p Params) String() string

type SpeechRecognition 

type SpeechRecognition interface {
	Start() error
	SpeechToText(data []byte) error
	Result() (string, error)
	Close() error
}

type SpeechRecognitionStandalone 

type SpeechRecognitionStandalone struct {
	SpeechRecognitionStream
}

func NewAutoDetectedSpeechRecognitionStandalone 

func NewAutoDetectedSpeechRecognitionStandalone(key, region string) (*SpeechRecognitionStandalone, error)

func NewSpeechRecognitionStandalone 

func NewSpeechRecognitionStandalone(key, region string, languages []string) (*SpeechRecognitionStandalone, error)

func (*SpeechRecognitionStandalone) SpeechToText 

func (s *SpeechRecognitionStandalone) SpeechToText(data []byte) error

type SpeechRecognitionStream 

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

func NewAutoDetectedSpeechRecognitionStream 

func NewAutoDetectedSpeechRecognitionStream(key, region string) (*SpeechRecognitionStream, error)

func NewSpeechRecognitionStream 

func NewSpeechRecognitionStream(key, region string, languages []string) (*SpeechRecognitionStream, error)

func (*SpeechRecognitionStream) Close 

func (s *SpeechRecognitionStream) Close() error

func (*SpeechRecognitionStream) Result 

func (s *SpeechRecognitionStream) Result() (string, error)

func (*SpeechRecognitionStream) SpeechToText 

func (s *SpeechRecognitionStream) SpeechToText(data []byte) error

func (*SpeechRecognitionStream) Start 

func (s *SpeechRecognitionStream) Start() error

type SpeechSynthesis 

type SpeechSynthesis interface {
	TextToSpeech(text string) error
	Result() (*WordBoundery, []byte, error)
	Error() error
	Close() error
}

type SpeechSynthesisStandalone 

type SpeechSynthesisStandalone struct {
	SpeechSynthesisStream
}

func NewAutoDetectedSpeechSynthesisStandalone 

func NewAutoDetectedSpeechSynthesisStandalone(key, region string) (*SpeechSynthesisStandalone, error)

func NewSpeechSynthesisStandalone 

func NewSpeechSynthesisStandalone(key, region string, voice string) (*SpeechSynthesisStandalone, error)

type SpeechSynthesisStream 

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

func NewAutoDetectedSpeechSynthesisStream 

func NewAutoDetectedSpeechSynthesisStream(key, region string) (*SpeechSynthesisStream, error)

func NewSpeechSynthesisStream 

func NewSpeechSynthesisStream(key, region string, voice string) (*SpeechSynthesisStream, error)

func (*SpeechSynthesisStream) Close 

func (s *SpeechSynthesisStream) Close()

func (*SpeechSynthesisStream) Error 

func (s *SpeechSynthesisStream) Error() error

func (*SpeechSynthesisStream) GetAllSupportLanguage 

func (s *SpeechSynthesisStream) GetAllSupportLanguage() map[string]bool

func (*SpeechSynthesisStream) GetAllSupportVoicesForLanguage 

func (s *SpeechSynthesisStream) GetAllSupportVoicesForLanguage(language string) []*Voice

func (*SpeechSynthesisStream) Result 

func (s *SpeechSynthesisStream) Result() (*WordBoundery, []byte, error)

func (*SpeechSynthesisStream) TextToSpeech 

func (s *SpeechSynthesisStream) TextToSpeech(text string) error

type StreamResult 

type StreamResult struct {
	Err error
	// contains filtered or unexported fields
}

func (*StreamResult) Next 

func (s *StreamResult) Next() string

type StreamSplitter 

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

func NewStreamSplitter 

func NewStreamSplitter(stream *StreamResult) StreamSplitter

func (StreamSplitter) Segment 

func (ss StreamSplitter) Segment(echo bool) string

type Voice 

type Voice struct {
	Name      string
	LocalName string
	Gender    string
	StyleList []string
}

func (*Voice) String 

func (v *Voice) String() string

type WordBoundery 

type WordBoundery struct {
	BounderyType common.SpeechSynthesisBoundaryType
	AudioOffset  float64
	Duration     float64
	TextOffset   uint
	WordLength   uint
	Text         string
}

func (WordBoundery) String 

func (w WordBoundery) String() string

Source Files

View all Source files

Directories

Path Synopsis

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.