Audio

Configure and manage voice input/output for your agent.

Overview

AI Dev Kit supports:

• Input Audio Recorder - Record user speech

• Output Audio Player - Play agent speech

• Transcription - Speech-to-text

• Text-to-Speech - Generate speech from text

Input Audio

Recording

[SerializeField] private InputAudioRecorder recorder;

// Start recording
recorder.StartRecording();

// Stop and get clip
AudioClip clip = recorder.StopRecording();

// Send to agent
await agent.SendAsync(clip);

Automatic Recording

// Agent handles recording automatically
await agent.SendAudioAsync();

Events

agent.onInputAudioStarted.AddListener(() => {
    ShowRecordingIndicator();
});

agent.onInputAudioCompleted.AddListener((clip) => {
    HideRecordingIndicator();
    Debug.Log($"Recorded: {clip.length}s");
});

agent.onInputAudioTranscribed.AddListener((text) => {
    Debug.Log($"You said: {text}");
});

Output Audio

Playback

[SerializeField] private OutputAudioPlayer player;

// Agent generates and plays audio automatically
await agent.SendAsync("Hello!");
// Speech plays automatically if EnableOutputAudio = true

Manual Playback

// Generate audio
AudioClip clip = await agent.GenerateSpeechAsync("Hello!");

// Play manually
player.Play(clip);

Events

agent.onOutputAudioStarted.AddListener(() => {
    ShowSpeakerIndicator();
});

agent.onOutputAudioCompleted.AddListener(() => {
    HideSpeakerIndicator();
});

agent.onOutputAudioProgress.AddListener((progress) => {
    UpdateProgressBar(progress);
});

Transcription

Configure

settings.InputAudioParameters = new TranscriptionParameters
{
    Model = "whisper-1",
    SpokenLanguage = SystemLanguage.English,
    Prompt = "Technical support conversation",
    Temperature = 0.2f
};

Manual Transcription

AudioClip clip = GetAudioClip();
string text = await agent.TranscribeAsync(clip);
Debug.Log($"Transcribed: {text}");

Text-to-Speech

Configure

settings.OutputAudioParameters = new SpeechParameters
{
    Model = "tts-1-hd",
    Voice = "nova",
    Speed = 1.0f,
    Format = AudioFormat.mp3
};

Manual TTS

AudioClip clip = await agent.GenerateSpeechAsync("Hello!");
audioSource.PlayOneShot(clip);

Volume Control

// Set volume
agent.OutputAudioVolume = 0.8f;

// UI slider
public void OnVolumeChanged(float value)
{
    agent.OutputAudioVolume = value;
}

Complete Example

using UnityEngine;
using Glitch9.AIDevKit.Agents;

public class VoiceChat : MonoBehaviour
{
    [SerializeField] private AgentBehaviour agent;
    [SerializeField] private InputAudioRecorder recorder;
    [SerializeField] private OutputAudioPlayer player;
    
    void Start()
    {
        // Setup
        agent.InputAudioRecorder = recorder;
        agent.OutputAudioPlayer = player;
        
        // Subscribe to events
        agent.onInputAudioStarted.AddListener(OnRecordingStarted);
        agent.onInputAudioCompleted.AddListener(OnRecordingCompleted);
        agent.onInputAudioTranscribed.AddListener(OnTranscribed);
        agent.onOutputAudioStarted.AddListener(OnPlaybackStarted);
        agent.onOutputAudioCompleted.AddListener(OnPlaybackCompleted);
    }
    
    public async void OnMicButtonPressed()
    {
        // Record and send
        await agent.SendAudioAsync();
    }
    
    void OnRecordingStarted()
    {
        Debug.Log("Recording...");
    }
    
    void OnRecordingCompleted(AudioClip clip)
    {
        Debug.Log($"Recorded {clip.length}s");
    }
    
    void OnTranscribed(string text)
    {
        Debug.Log($"You: {text}");
    }
    
    void OnPlaybackStarted()
    {
        Debug.Log("Playing response...");
    }
    
    void OnPlaybackCompleted()
    {
        Debug.Log("Playback complete");
    }
}

Next Steps

• Input Audio Recorder

• Output Audio Player

• Transcription

• Text-to-Speech

• Audio Setup