Omni-OpenAI

OpenAI provider adapters for the omni-* ecosystem, wrapping the official openai-go SDK.

Features

• OmniLLM: Chat completions provider with streaming, tool calling, and vision support
• OmniVoice STT: Whisper transcription with word and segment timestamps
• OmniVoice TTS: OpenAI audio synthesis with multiple voices

Installation

go get github.com/plexusone/omni-openai

Usage

OmniLLM Provider

package main

import (
    "context"
    "log"
    "os"

    core "github.com/plexusone/omnillm-core"
    _ "github.com/plexusone/omni-openai/omnillm" // Auto-registers provider
)

func main() {
    ctx := context.Background()

    // Create provider via registry
    provider, err := core.NewProvider("openai", core.ProviderConfig{
        APIKey: os.Getenv("OPENAI_API_KEY"),
    })
    if err != nil {
        log.Fatal(err)
    }
    defer provider.Close()

    resp, err := provider.CreateChatCompletion(ctx, &core.ChatCompletionRequest{
        Model: "gpt-4o",
        Messages: []core.Message{
            {Role: core.RoleUser, Content: "Hello!"},
        },
    })
    if err != nil {
        log.Fatal(err)
    }
    log.Printf("Response: %s", resp.Choices[0].Message.Content)
}

OmniVoice STT Provider

package main

import (
    "context"
    "log"

    "github.com/plexusone/omnivoice-core/stt"
    "github.com/plexusone/omni-openai/omnivoice"
)

func main() {
    ctx := context.Background()

    // Create STT provider
    provider, err := omnivoice.NewSTTProviderFromEnv()
    if err != nil {
        log.Fatal(err)
    }

    result, err := provider.Transcribe(ctx, audioData, stt.TranscriptionConfig{
        Encoding: "mp3",
    })
    if err != nil {
        log.Fatal(err)
    }
    log.Printf("Transcription: %s", result.Text)
}

OmniVoice TTS Provider

package main

import (
    "context"
    "log"

    "github.com/plexusone/omnivoice-core/tts"
    "github.com/plexusone/omni-openai/omnivoice"
)

func main() {
    ctx := context.Background()

    // Create TTS provider
    provider, err := omnivoice.NewTTSProviderFromEnv()
    if err != nil {
        log.Fatal(err)
    }

    result, err := provider.Synthesize(ctx, "Hello, world!", tts.SynthesisConfig{
        VoiceID: omnivoice.VoiceNova,
    })
    if err != nil {
        log.Fatal(err)
    }
    // result.Audio contains the MP3 audio data
}

Direct OpenAI Client Usage

package main

import (
    "context"
    "log"

    openai "github.com/plexusone/omni-openai"
)

func main() {
    client, err := openai.NewClientFromEnv()
    if err != nil {
        log.Fatal(err)
    }

    ctx := context.Background()

    // Transcribe audio
    resp, err := client.Transcribe(ctx, openai.TranscriptionRequest{
        Audio:    audioData,
        Filename: "audio.mp3",
    })
    if err != nil {
        log.Fatal(err)
    }
    log.Printf("Transcription: %s", resp.Text)

    // Synthesize speech
    ttsResp, err := client.Synthesize(ctx, openai.TTSRequest{
        Input: "Hello, world!",
        Voice: openai.VoiceAlloy,
    })
    if err != nil {
        log.Fatal(err)
    }
    // ttsResp.Audio contains the MP3 audio data
}

Configuration

Set the OPENAI_API_KEY environment variable or pass the API key directly:

client := openai.NewClient("your-api-key")

Available TTS Voices

Voice	Description
alloy	Neutral, balanced
ash	Warm, engaging
ballad	Melodic, expressive
coral	Clear, articulate
echo	Smooth, natural
fable	Storytelling, dramatic
nova	Bright, energetic
onyx	Deep, resonant
sage	Calm, wise
shimmer	Light, cheerful
verse	Poetic, rhythmic
marin	Friendly, approachable
cedar	Grounded, trustworthy

Package Structure

omni-openai/
├── openai.go           # Direct OpenAI client (STT/TTS)
├── omnillm/            # OmniLLM provider adapter
│   └── adapter.go
└── omnivoice/          # OmniVoice provider adapters
    ├── stt.go
    └── tts.go

License

MIT License - see LICENSE for details.