PDF轉音訊工具

A PDF to audio converter extracts text from a PDF document and converts it into spoken audio using text-to-speech (TTS) technology. WhiskFile uses the Kokoro 82M neural TTS model running directly in your browser for natural-sounding speech.

Upload your PDF, select a voice from the 11 available options, optionally edit the extracted text, and click Synthesize. The audio is generated sentence by sentence in your browser. Download the result as a WAV file.

No. The Kokoro TTS model runs 100% in your browser using WebAssembly. Your PDF and the generated audio never leave your device. This is fundamentally more private than cloud-based TTS services.

The Kokoro 82M model produces very natural-sounding speech with proper intonation, rhythm, and emphasis. It is a neural TTS model trained on high-quality speech data — far more natural than robotic rule-based TTS.

11 voices are available: 7 American English voices (Heart, Bella, Sarah, Nicole, Sky — female; Adam, Michael — male) and 4 British English voices (Emma, Isabella — female; George, Lewis — male). Each voice has a distinct speaking style.

You can download in MP3 or WAV format. MP3 (128 kbps) is compact and ideal for phones, music players, and portable listening. WAV (24 kHz, lossless) preserves full audio quality. Both are generated directly in your browser.

Yes. After generating the audio, click Download MP3 to get a standard 128 kbps MP3 file. The MP3 is encoded entirely in your browser — your audio never leaves your device. You can also download as WAV for lossless quality.

Yes. After the PDF text is extracted, you can edit it in the text editor. Remove headers, footers, page numbers, or any unwanted content before synthesis.

Yes. PDF to Audio is completely free with no signup, no daily limits, and no watermarks. The Kokoro TTS model runs in your browser at zero cost.

Yes. Use the speed control in the audio player to adjust playback between 0.5x and 2x speed. This is useful for studying (slower) or quickly reviewing long documents (faster).

Synthesis speed depends on your device and the text length. The model processes text sentence by sentence. A 10-page PDF typically takes 1-3 minutes on a modern computer. Progress is shown in real time.

Yes. After synthesis, the audio plays directly in the built-in waveform player. You can listen, pause, seek, and navigate by sentences — all in your browser without downloading anything.

Yes. The PDF to Audio converter works on iPhones, Android devices, and tablets. However, neural TTS synthesis is CPU-intensive, so it may be slower on older mobile devices.

The maximum file size is 100 MB and the maximum page count is 200 pages. Very long documents (over 200,000 characters) are truncated. For longer documents, consider splitting the PDF first.

Yes. Upload your PDF book, select a natural-sounding voice, and generate audio for the entire text. While it does not detect chapters automatically, you can edit the text before synthesis to organize the content.

Yes. Upload a textbook PDF with selectable text and the tool will extract and synthesize all text content. This is ideal for students who want to listen to study material during commutes or exercise.

Not directly. This tool extracts text from PDF documents with selectable text. If your PDF is a scanned image, use the OCR PDF tool first to make the text extractable, then use PDF to Audio.

No. Password-protected or encrypted PDFs must be unlocked first. Use WhiskFile's Unlock PDF tool to remove the password, then upload the unlocked PDF for audio conversion.

The Kokoro 82M model is approximately 86 MB (quantized). It downloads once from Hugging Face's CDN and is cached in your browser for instant reuse on subsequent visits.

The Kokoro TTS model currently supports English with American and British accents. For PDFs in other languages, text will be extracted but the speech synthesis produces English pronunciation.

Use text-based PDFs (not scanned images) for the best results. Edit the extracted text before synthesis to remove headers, footers, page numbers, and formatting artifacts. Simple document layouts produce better speech than complex multi-column layouts or tables.

The tool works best with text-based PDFs in English. Scanned PDFs require OCR first. Complex layouts (multi-column, tables, mathematical formulas) may produce imperfect text extraction. The TTS model handles standard prose well but may struggle with technical symbols, code, or non-English text. Maximum limits are 100 MB, 200 pages, and 200,000 characters.

是的。PDF轉音訊完全在您的瀏覽器中使用WebAssembly運行。您的檔案不會上傳到任何伺服器——整個過程中始終保留在您的裝置上。