Chuyển đổi âm thanh sang PDF

An audio to PDF converter transcribes spoken audio into text using speech recognition, then formats the resulting transcript into a downloadable, searchable PDF document. WhiskFile uses OpenAI's Whisper AI model running directly in your browser to perform the transcription privately.

Upload your audio file (MP3, WAV, M4A, etc.), select a Whisper model and language, click Transcribe, review the transcript, customize the PDF layout, and download. The entire process runs in your browser — no upload to any server.

Accuracy depends on the model size and audio quality. Small EN (~244 MB) offers the best accuracy for English. Large v3 Turbo (~809 MB) gives maximum accuracy across all languages. Clear audio with minimal background noise produces the best results.

No. The Whisper AI model runs 100% in your browser using WebAssembly. Your audio files never leave your device. This is fundamentally more private than cloud-based transcription services.

Yes. Upload your MP3 file, transcribe it with Whisper AI, and download the transcript as a searchable PDF. WhiskFile supports MP3 along with WAV, M4A, OGG, FLAC, WebM, AAC, and Opus.

Yes. WAV files are fully supported. Upload your WAV recording, transcribe it, and export the result as a professionally formatted PDF transcript.

Yes. M4A files (common on iPhones and Apple devices) are supported. Upload the M4A file and WhiskFile will transcribe the audio and generate a PDF transcript.

Yes. Upload any voice recording in a supported format (MP3, WAV, M4A, OGG, FLAC, WebM, AAC, Opus) or record directly from your microphone. WhiskFile transcribes the speech and generates a formatted PDF.

Yes. Upload your meeting recording from Zoom, Google Meet, Teams, or any other platform. WhiskFile will transcribe the audio and produce a searchable PDF of the meeting minutes.

Yes. Upload lecture recordings to create searchable PDF study notes. Use word-level timestamps to reference specific moments, and choose the Transcript layout for easy reading.

MP3, WAV, M4A (AAC), OGG (Vorbis), FLAC, WebM, AAC, and Opus. These cover the most common audio formats from phones, recorders, and online platforms. The browser's Web Audio API handles decoding.

Yes. Multilingual models support 20 languages including English, Hindi, Spanish, French, German, Italian, Portuguese, Chinese, Japanese, Korean, Arabic, Russian, Dutch, Turkish, Swedish, Polish, Thai, Vietnamese, and Indonesian, plus automatic language detection.

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

No. The audio to PDF converter works without any account or signup. Just open the page, upload your audio, and start transcribing.

The maximum file size is 500 MB. There is no hard duration limit, but very long audio files (over 30 minutes) may take several minutes to process depending on your device and the model size. The estimated transcription time is displayed before you start.

Yes. After transcription, you can edit the full text directly in the editor. When timestamps are disabled, the transcript appears as an editable text area where you can correct errors, add formatting, or remove content before exporting.

Yes. Toggle timestamps on or off in the PDF Export Options panel. When enabled, each segment includes its timecode in the PDF. The Transcript layout style places timestamps on their own line for easy reading.

Yes. In addition to PDF, you can export as plain text (TXT) with or without timestamps, SRT subtitle format, copy the full text to clipboard, or print directly from the browser.

Yes. The audio to PDF converter works on iPhone, Android, iPad, and tablets. Upload files from your device storage or record from the microphone. The interface adapts to mobile screen sizes.

Whisper models range from ~75 MB (Tiny) to ~809 MB (Large v3 Turbo). They download once from Hugging Face's CDN and are cached in your browser for instant reuse on subsequent visits.

Phrase-level timestamps group words into natural segments (sentences). Word-level timestamps assign a timecode to each individual word — useful for precise subtitle sync. Choose your preferred mode before transcribing.

Speaker diarization (identifying who said what) is not currently supported. Whisper transcribes all speech as a single speaker. For multi-speaker recordings, you can manually label speakers in the editable transcript after transcription.

Three styles: Minimal (clean, no header), Professional (title, date, language, divider line), and Transcript (timestamps on their own line). You can also choose page size (A4, Letter, Legal), orientation (Portrait, Landscape), adjustable font size (9–14pt), and optional page numbers.

Có. Âm thanh sang PDF chạy hoàn toàn trong trình duyệt bằng WebAssembly. Tệp của bạn không bao giờ được tải lên máy chủ — chúng ở trên thiết bị của bạn trong suốt quá trình.