Switching recording off does not affect the transcript. Speech recognition runs independently, so you still get the full transcript, summary, sentiment and extracted data - just no audio file.
Who records
When recording is on, you choose who captures the audio. The default is your carrier - exactly what happened before this setting existed.
Set it per agent in the dashboard (Call recording -> Recorded by) or with
recording_source on Create agent and Update agent. A single number can override its agent with recording_source on Update phone number; null on a number means inherit the agent. Any other value is rejected with invalid_recording_source.
Recording off always wins. With
record_call off nothing is recorded by anyone, whatever recording_source says.Turning it off
Per agent (dashboard) Open the agent, find Call actions, and switch off Call recording. Every number on that agent stops recording from the next call onwards. Per number (API) A single number can override its agent - useful when one client’s privacy policy forbids recording but the rest of your numbers are fine.
Agent-level can also be set through the API with
record_call on Create agent and Update agent.
What happens when it is off
- No audio is captured at all. Nixflex never asks the carrier to record, so there is no file anywhere - not on your carrier, not with us.
recording_urlisnullon the call record and in thecall.completedwebhook. If your code reads that field, handlenull.- Past recordings are untouched. Switching off applies to future calls only.
- You are not charged by your carrier for recording those calls.
Where the audio is stored
When a call ends, a carrier recording is downloaded from your carrier, stored in EU (Ireland) storage, and the carrier’s copy is deleted. A Nixflex recording is written straight to that same EU storage - there is never a carrier copy to fetch or delete. Prefer the audio in your own region? Connect your own S3-compatible bucket and recordings go there instead, with Nixflex keeping no copy - see Your own storage.This is what both carriers recommend. Twilio charges for recording storage and documents downloading recordings to avoid those charges. Telnyx keeps recordings for up to a year and advises fetching what you need in good time. Pulling the file to EU storage also keeps recordings inside the EU, which matters if your end-customers are covered by GDPR.
Reading a recording
The URL arrives in two places:- On the call record:
GET /v1/calls/:idreturnsrecording_url - In the post-call webhook:
call.completedincludesrecording_url
Audio format
Dual-channel audio is easier to analyse per speaker. The carrier difference between inbound and outbound is long-standing behaviour and is kept as-is so existing integrations are not changed. If your code assumes a file extension, read it from
recording_url rather than building the name yourself.
Consent
Testing it
1
Switch it off
Turn off Call recording on the agent, or PATCH
record_call: false on the number.2
Make a call
Call the number and talk for a few seconds, then hang up.
3
Check the call record
GET /v1/calls/:id - recording_url should be null, while the transcript and summary are still there.4
Switch it back on
Set it back to
true, or null on the number to inherit the agent again, and call once more. recording_url should be populated.Related
- Update phone number - the
record_callfield - Update phone number - the
recording_sourcefield - Data retention - how long recordings are kept
- Webhooks -
recording_urlincall.completed - Keypad input - the other per-number call setting