LISTEN & LOCATE
Find the moment behind a word.
Follow the recording and jump back to a segment. Word timing connects the text to the audio so the important details are easy to check.
WhisperX online.
Your words, ready.
Upload audio or video. Edit and export your transcript. No local setup.
Independent, unofficial hosted service
Your personal transcription workspace
Powered by WhisperX
SAMPLE RESULT
Quick start
Play it. Change a word. Make it yours.
Sample changes stay here
00:0000:15
Result details & original output
WhisperX · large-v3 · ENOriginal demonstration script; eSpeak NG synthetic voice; actual WhisperX large-v3 output.
Real WhisperX processing Word-level output All four export formats Pay as you go
A WORKSPACE, NOT AN INSTALLATION PROJECT
From a recording to something useful.
WhisperX brings transcription, alignment, and optional speaker diarization together. We make that pipeline available in a workspace you can start using right away.
SPEAKER LABELS
Give the conversation some shape.
Group voice turns, rename the speakers, and correct individual labels. Speaker groups describe this recording; they do not identify people.
EDIT NATURALLY
A little polish goes a long way.
Fix a name or a sentence right where you read it. Your changes save automatically, with the original model result always available.
TAKE IT WITH YOU
The format your next step needs.
Readable text, structured JSON, SRT subtitles, or web-ready VTT. Download an existing result again without using extra processing minutes.
LOOK INSIDE THE RESULT
Timing you can inspect. Output you can keep.
A transcript is more useful when you can check where its data came from. Our public samples include the input recording and the actual model output.
ACTUAL SAMPLE · WORD-LEVEL JSONlarge-v3
{
"word": "Welcome",
"start": 0.051,
"end": 0.492,
"timingStatus": "unknown"
}Download the complete sample The original is always a reference.
WhisperX predicts text, then aligns it to the recording. Some timestamps may be missing or estimated. A text correction keeps the sentence timing, but does not measure fresh word timing. Keep the original JSON when that distinction matters to your workflow.
TXTJSONSRTVTT
ONE ENGINE, YOUR WORKFLOW
Start with what you want to make.
Choose a focused workflow. Each tool uses the same real WhisperX engine, with the right starting options and export guidance.
ONE ENGINE. YOUR WORKFLOW.
Take your recording somewhere useful.
YOU HAVE OPTIONS
WhisperX, with or without the setup.
The open source project is available to run locally. Our hosted service is for people who want the output without managing the processing environment.
Run it here.
Upload your file, confirm the processing minutes, and get your result in your account. We handle model access and the hosted environment.
Try the editorRun it your way.
Use the project locally when you want control of your environment and configuration. CPU mode is an option; a compatible GPU can accelerate processing.
Read the installation guideWhisperX.app is not affiliated with or endorsed by the WhisperX authors or OpenAI. Visit the upstream project.
MINUTES, NOT COMMITMENTS
Use it when you need it.
Start with a five-minute trial. Buy a pack for your next project. Transcription, speaker labels, and all export formats are included.
Free trial
Get a feel for the workflow
$0USD
5 minutes
- 5 minutes, once per verified account
- All four export formats
- Speaker labels included
- Results available for 7 days
No card required. Trial minutes expire in 30 days.
Starter
For an interview or a project
$9USD once
120 minutes
- 120 minutes to use as needed
- Audio and video transcription
- Text and speaker editing
- Results available for 30 days
One payment. No automatic renewal.
Creator
For recordings that keep coming
$29USD once
600 minutes
- 600 minutes to use as needed
- Everything in Starter
- The same model and export formats
- Lower price per minute
One payment. No automatic renewal.
A LITTLE HELP, WHEN YOU WANT IT
Understand the pipeline. Get better results.
Practical guides to using WhisperX, reviewing speaker labels, and choosing between a local environment and hosted processing.
GET YOUR BEARINGS
Understand the tool you are using.
WHISPERX GUIDE
How to use WhisperX ↗
The shortest route from a recording to a WhisperX result is to upload, run, and download. Use the editor when the recording needs a little more attention.
WHISPERX GUIDELocal installation ↗
If you want control of the environment, run the open source project locally. This guide explains the setup decisions and links to the current upstream instructions.
WHISPERX GUIDEWhisperX vs Whisper ↗
WhisperX builds a broader transcription pipeline around speech recognition. Its additional stages are most useful when your output needs timing and speaker information.
WHISPERX GUIDESpeaker label guide ↗
A diarized transcript is easier to read when labels are checked and named. Start with clear turns, then examine the ambiguous parts of the conversation.
GOOD TO KNOW
Before you press play.
Is this the official WhisperX website?
WhisperX.app is an independent, unofficial hosted service. The open source project is maintained separately at m-bain/whisperX on GitHub.
Do I need a GPU or a model token?
No. The hosted service handles the processing environment. Upload your own file, confirm the charge, and run the task from your account.
What do I get back?
An editable transcript, timing information, optional speaker labels, and downloads in TXT, JSON, SRT, or VTT. Review important words and speaker assignments before using the result.
Does editing cost more?
Editing an existing result and downloading it again do not use more minutes. A new transcription run is a new task and requires a new confirmed charge.
Are word timestamps always available?
Some words cannot be aligned reliably. Missing or uncertain timing is kept distinct from measured timing. Editing a sentence does not automatically align the new words.
Can I transcribe a link?
Use a file upload. The service does not fetch YouTube videos, external recordings, or other media links.
LESS SETUP. MORE DONE.
Your next transcript starts here.
Try the public sample, then bring your own recording. Your first account includes 5 minutes to try the workflow.
Run WhisperX