Export to m4b from WAV efficiently

Author: richardr1126

src/app/api/audio/convert/route.ts

@@ -87,26 +87,16 @@ export async function POST(request: NextRequest) {
       await mkdir(intermediateDir);
     }
 
-    // First, write each chapter to a temporary file and get its duration
+    // Process each chapter - no need for initial conversion since input is WAV
     const chapterFiles: { path: string; title: string; duration: number }[] = [];
     let currentTime = 0;
 
     for (let i = 0; i < data.chapters.length; i++) {
       const chapter = data.chapters[i];
-      const inputPath = join(intermediateDir, `${i}-input.aac`);
       const outputPath = join(intermediateDir, `${i}.wav`);
 
-      // Write the chapter audio to a temp file
-      await writeFile(inputPath, Buffer.from(new Uint8Array(chapter.buffer)));
-      
-      // Convert to WAV with consistent format (this helps with timestamp issues)
-      await runFFmpeg([
-        '-i', inputPath,
-        '-acodec', 'pcm_s16le',
-        '-ar', '44100',
-        '-ac', '2',
-        outputPath
-      ]);
+      // Write the chapter audio directly since it's already WAV
+      await writeFile(outputPath, Buffer.from(new Uint8Array(chapter.buffer)));
 
       // Get the duration of this chapter
       const duration = await getAudioDuration(outputPath);
@@ -116,9 +106,6 @@ export async function POST(request: NextRequest) {
         title: chapter.title,
         duration
       });
-
-      // Clean up input file
-      await unlink(inputPath).catch(console.error);
     }
 
     // Create chapter metadata file
@@ -129,7 +116,7 @@ export async function POST(request: NextRequest) {
     );
 
     // Calculate chapter timings based on actual durations
-    chapterFiles.forEach((chapter, index) => {
+    chapterFiles.forEach((chapter) => {
       const startMs = Math.floor(currentTime * 1000);
       currentTime += chapter.duration;
       const endMs = Math.floor(currentTime * 1000);

src/app/api/tts/route.ts

@@ -29,15 +29,16 @@ export async function POST(req: NextRequest) {
       voice: voice as "alloy",
       input: text,
       speed: speed,
-      response_format: format === 'aac' ? 'aac' : 'mp3',
+      // Use wav format for audiobook generation to avoid initial conversion
+      response_format: format === 'audiobook' ? 'wav' : (format === 'aac' ? 'aac' : 'mp3'),
     }, { signal: req.signal }); // Pass the abort signal to OpenAI client
 
     // Get the audio data as array buffer
     // This will also be aborted if the client cancels
     const stream = response.body;
 
     // Return audio data with appropriate headers
-    const contentType = format === 'aac' ? 'audio/aac' : 'audio/mpeg';
+    const contentType = format === 'audiobook' ? 'audio/wav' : (format === 'aac' ? 'audio/aac' : 'audio/mpeg');
     return new NextResponse(stream, {
       headers: {
         'Content-Type': contentType

src/contexts/EPUBContext.tsx

@@ -224,7 +224,7 @@ export function EPUBProvider({ children }: { children: ReactNode }) {
               text: text.trim(),
               voice: voice,
               speed: voiceSpeed,
-              format: 'aac'
+              format: 'audiobook'  // Request WAV format directly
             }),
             signal
           });
@@ -244,7 +244,7 @@ export function EPUBProvider({ children }: { children: ReactNode }) {
             let spineIndex = processedSections;
             let currentSpineHref: string | undefined;
 
-            spine.each((item: any) => {
+            spine.each((item: SpineItem) => {
               if (spineIndex === 0) {
                 currentSpineHref = item.href;
               }

src/contexts/PDFContext.tsx

@@ -36,7 +36,6 @@ import {
 } from '@/utils/pdf';
 
 import type { PDFDocumentProxy } from 'pdfjs-dist';
-import { useParams } from 'next/navigation';
 
 /**
  * Interface defining all available methods and properties in the PDF context
@@ -234,7 +233,7 @@ export function PDFProvider({ children }: { children: ReactNode }) {
               text: text.trim(),
               voice: voice,
               speed: voiceSpeed,
-              format: 'aac'
+              format: 'audiobook'  // Request WAV format directly
             }),
             signal
           });