muninn/app/Support/VideoWorkflow/StepHandlers/TtsVoiceoverHandler.php

<?php

namespace App\Support\VideoWorkflow\StepHandlers;

use App\Models\AiProvider;
use App\Models\WorkflowStep;
use App\Support\AiProviders\AiGenerationException;
use App\Support\AiProviders\AiProviderManager;
use App\Support\AssetStorage\AssetStorage;
use App\Support\VideoWorkflow\WorkflowEngine;

/**
 * Synthesizes narration audio via an ElevenLabs-backed AiProvider. Text
 * comes from $step->input['text'] if given, otherwise the project's latest
 * 'script' asset. Voice selection (input.voice_id / input.model_id) is
 * per-step, falling back to the AiProvider's own config.
 *
 * Note: ElevenLabs' base TTS endpoint doesn't expose a direct "speed"
 * parameter — only voice_id/model_id are wired through here for now.
 */
class TtsVoiceoverHandler implements StepHandler
{
    public function __construct(
        private readonly AiProviderManager $aiProviders,
        private readonly AssetStorage $assets,
    ) {}

    public function type(): string
    {
        return 'tts_voiceover';
    }

    public function handle(WorkflowStep $step, WorkflowEngine $engine): void
    {
        $provider = AiProvider::query()->where('provider', 'elevenlabs')->where('is_active', true)->first();

        if (! $provider) {
            $engine->failStep($step, 'No active ElevenLabs provider configured.');

            return;
        }

        $text = $this->resolveText($step);

        if (! $text) {
            $engine->failStep($step, 'No script text available to voice — set input.text or complete a script_generation step first.');

            return;
        }

        $input = $step->input ?? [];
        $options = array_filter([
            'voice_id' => $input['voice_id'] ?? null,
            'model_id' => $input['model_id'] ?? null,
        ]);

        try {
            $result = $this->aiProviders->generate($provider, $text, $options);
        } catch (AiGenerationException $e) {
            $engine->failStep($step, $e->getMessage());

            return;
        }

        $asset = $this->assets->store(
            $step->videoProject,
            'audio',
            $result->content,
            'mp3',
            step: $step,
            mimeType: $result->contentType,
        );

        $engine->completeStep($step, [
            'asset_id' => $asset->id,
            'characters' => strlen($text),
        ]);
    }

    private function resolveText(WorkflowStep $step): ?string
    {
        $input = $step->input ?? [];

        if (! empty($input['text'])) {
            return $input['text'];
        }

        $scriptAsset = $step->videoProject->assets()
            ->where('type', 'script')
            ->orderByDesc('version')
            ->first();

        return $scriptAsset ? $this->assets->contents($scriptAsset) : null;
    }
}