perf: skip ffmpeg transcode for PocketTTS streaming — use WAV directly

PocketTTS emits 16-bit PCM WAV (16 kHz mono). WAV is natively supported
on all browsers including iOS/macOS Safari, so the ffmpeg MP3 transcode
is unnecessary for the streaming path.

Using format=wav for PocketTTS voices eliminates the ffmpeg subprocess
startup delay (~200–400 ms) and a pipeline stage, giving lower latency
to first audio frame. Kokoro and CF AI continue using MP3 (they output
MP3 natively or via the OpenAI-compatible endpoint).

The runner (MinIO storage) is unaffected — it still stores MP3 for
space efficiency.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

ui/src/lib/components/AudioPlayer.svelte

@@ -574,7 +574,12 @@
 			// generation runs and MinIO is populated concurrently.
 			// Skip when enqueued=true to avoid double-generation with the async runner.
 			if (!voice.startsWith('cfai:') && !presignResult.enqueued) {
-				const qs = new URLSearchParams({ voice, format: 'mp3' });
+				// PocketTTS outputs raw WAV — skip the ffmpeg transcode entirely.
+				// WAV (PCM) is natively supported on all platforms including iOS Safari.
+				// Kokoro and CF AI output MP3 natively, so keep mp3 for those.
+				const isPocketTTS = voices.some((v) => v.id === voice && v.engine === 'pocket-tts');
+				const format = isPocketTTS ? 'wav' : 'mp3';
+				const qs = new URLSearchParams({ voice, format });
 				const streamUrl = `/api/audio-stream/${slug}/${chapter}?${qs}`;
 				// HEAD probe: check paywall without triggering generation.
 				const headRes = await fetch(streamUrl, { method: 'HEAD' }).catch(() => null);