Recognition task

Audio captcha API

Send a spoken-digit audio challenge as base64 and get the transcription back as text, usually in about a second.

Price / 1,000
—
Typical speed
—
Success
—
Get API key Read the docs

Pay only for solved tasks. Failed tasks are refunded automatically.

What you send, what you get

What is Audio captcha?

Audio captchas play a short clip of spoken digits or letters, often with background noise. They exist as the accessible alternative to image challenges. bypass.so returns what is spoken as a plain string.

Recognition tasks usually finish inside the createTask call, so the first response already carries the answer. Typical time to a result is about a second.

POST/createTask
GET/getTaskResult/{taskId}
POST/reportBad?task_id={taskId}
Every request needs your API key. Create an account to get one, top up any amount, and you are ready to send tasks.

Request parameters

FieldDescription
clientKeystringrequiredYour API key. You can also send it as Authorization: Bearer <key> instead.
typestringrequiredAlways AudioCaptchaTask for this task.
audioBase64stringrequiredThe audio clip as base64, MP3 or WAV, without a data: prefix.
contentTypestringoptionalMIME type of the clip, for example audio/wav. Defaults to audio/mpeg.
Request · POST /createTask














Response fields


FieldMeaning


errorId0 on success, 1 on error. Errors also carry errorCode and errorDescription.


statusprocessing while the task runs, then ready or failed.


taskIdReturned by createTask. Use it to poll and to report a wrong answer.


solution.textThe spoken characters as a string, without spaces.



















Integration


Copy, paste, solve


Plain HTTPS and JSON. No SDK to install.


    

  1. Create the taskSend your key, AudioCaptchaTask and its fields.
    2. 

  2. Poll for the resultCheck every second while status is processing.
    3. 

  3. Use the textRead solution.text and continue your flow.
    4. 

  4. Wrong answer? Report itPOST /reportBad?task_id=<taskId> with your key as Bearer returns that task's price once.
    5. 




Python guide
Node.js guide
All integrations

















Errors and limits


errorCodeWhen it happens


ERROR_KEY_INVALIDThe API key is missing or not recognised (HTTP 403).


ERROR_INSUFFICIENT_CREDITSYour balance does not cover the task price. Top up and resend (HTTP 402).


ERROR_TASK_NOT_SUPPORTEDThe type value is unknown or not live yet (coming soon). Nothing is charged.


ERROR_BAD_REQUESTThe body is not valid JSON or a field has the wrong type (HTTP 400).


ERROR_CAPTCHA_UNSOLVABLEThe task could not be completed. The held price is refunded automatically.


ERROR_RATE_LIMITToo many requests in a short time. Back off and retry (HTTP 429).


ERROR_NOT_FOUNDUnknown taskId, or the task belongs to another key (HTTP 404).




Check your balance any time with GET /balance and Authorization: Bearer <key>. Full list in the error reference.

















Using the result


From text to submitted form


What your own script does with the transcribed text once the task is ready.




    

  1. Fetch the clip
    Download the audio file the challenge plays, in the same session that shows the challenge.
    2. 

  2. Send it as base64
    Create an AudioCaptchaTask with the clip in audioBase64; set contentType for WAV files.
    3. 

  3. Enter the digits
    Type solution.text into the response field and submit the form.
    4. 



Result fieldsolution.textSend the clip unchanged. Re-encoding or trimming the audio rarely helps and can cut off the first digit.













Use cases


Where teams use Audio captcha tasks


Built for testing, accessibility and data work on sites you own or are allowed to automate.




    

  • Accessibility tooling
    Build assistive flows for users with hearing or vision impairments who meet audio-only challenges.
    • 

  • Testing audio fallbacks
    Check that the accessible path of your own forms works end to end in CI, not only the visual one.
    • 

  • Contact-centre and data tools
    Automate permitted lookups on portals that offer the audio option.
    • 



Use bypass.so only where you have permission. Respect each site's terms and rate limits, and never use results to access accounts or data that are not yours. See our acceptable use policy.













Pricing


Audio captcha pricing


—per 1,000 solved tasks


That is — per task, or — tasks for $100.


    

  • No subscription, no minimum volume
    • 

  • Failed tasks are refunded automatically
    • 

  • Bigger top-ups get bonus credit
    • 



Full pricing and calculator






Volume bonus on a single top-up


TierTop-upBonusEffective / 1kTasks covered















Related types
Same key, more task types

All 7 types
















FAQ
Audio captcha questions
More answers in the general FAQ and the API reference.




Which audio formats are supported?
MP3 and WAV, sent as base64 in audioBase64. Set contentType to audio/wav for WAV; MP3 is the default.


What does the answer look like?
A plain string such as 480917, with no spaces or punctuation, ready to type into the response field.


How long can a clip be?
Short challenge clips of a few seconds work best. Send the file as the challenge provides it rather than a recording of your speakers.


What if a transcription is wrong?
Report it with POST /reportBad and the taskId. The task's price is returned to your balance once per task.


Do failed tasks cost anything?
No. When a clip cannot be transcribed the held price is refunded automatically.








Send your first Audio captcha task
Create an account, top up from a few dollars and get your key. You only pay for tasks that return a result.


Get API keyRead the docs