Short answer: use SRTP to protect the voice-media stream and TLS to protect SIP signaling on every configured signaling hop. Then verify key exchange, certificate validation, endpoints, administrator access, provider boundaries, recording systems, gateways, and monitoring. No single product name, firewall rule, VLAN, VPN, or caller-ID feature proves that an entire call is confidential.
Reviewed: August 18, 2026.
Source policy: this guide uses primary NIST, IETF RFC Editor, CISA, and FCC material. NIST SP 800-58 is a foundational VoIP security publication from 2005, so it is used for architecture and threat concepts, not as a current cipher-selection list. Protocol RFCs define technical boundaries, while current provider documentation, release notes, licenses, and the deployed configuration must confirm what is supported and enabled today.
Freshness rule: recheck the exact phones, soft clients, PBX or UCaaS platform, session border controller, carrier path, certificates, cryptographic profiles, recording integrations, and administrative controls before procurement and after material platform changes.
Start with the complete VoIP call path
VoIP separates call control from the audio stream. SIP commonly handles registration, call setup, routing, and teardown. RTP commonly carries the real-time audio. A call may cross a phone or soft client, local switches and Wi-Fi, a firewall, an SBC, a hosted provider, a carrier, a PSTN gateway, and the far endpoint. Recording, transcription, contact-center, CRM, and analytics services may create additional copies or termination points.
Draw that path before selecting controls. Mark where signaling is created, decrypted, inspected, and re-encrypted. Mark where media keys are established, which systems can access clear audio, where recordings are stored, and which organizations administer each component. Rivell’s VoIP phone service and business phone installation pages describe the related platform and deployment scope.
Which technology protects each part?
| Call-path area | Primary control | What to verify |
|---|---|---|
| Voice media | SRTP or a documented equivalent | Encryption and message authentication are active for every intended media leg, with an approved profile and no silent downgrade. |
| SIP signaling | TLS with validated certificates | Each required signaling hop uses TLS, peers validate the expected certificate identity, and insecure fallback is rejected where policy requires protection. |
| Media-key establishment | A supported key-management method, such as DTLS-SRTP where applicable | Keys are authenticated, scoped, rotated, stored safely, and not exposed in logs or insecure signaling. |
| Phones and soft clients | Supported firmware, device controls, and protected credentials | Default passwords are removed, updates are supported, local access is controlled, and lost devices can be disabled. |
| Administration | MFA, least privilege, restricted management paths, and audit logs | Provider, customer, installer, and support responsibilities are documented and tested. |
| Remote network segment | A securely configured VPN when the architecture requires one | The tunnel protects only its defined segment; media and signaling remain protected beyond the tunnel endpoint as required. |
| Recording and integrations | Access, encryption, retention, export, and deletion controls | Every intentional clear-audio or stored-copy boundary is approved and monitored. |
SRTP protects the media stream
The SRTP specification in RFC 3711 defines confidentiality for RTP and RTCP payloads, plus message authentication and replay protection when those services are used. It also makes an important boundary explicit: key management is external to SRTP. A platform must establish and protect the master key and cryptographic context correctly.
Do not accept a checkbox labeled “SRTP supported” as deployment evidence. Confirm that SRTP is active on the real call path, that the selected profile does not use null encryption, that message authentication and replay protection meet policy, and that calls fail or alert when a protected profile cannot be negotiated. RFC 3711 also notes that standard RTP header information can remain visible, so SRTP does not hide all traffic metadata.
TLS protects signaling hops, not the audio by itself
TLS 1.3 defines authentication, confidentiality, and integrity for the channel between its connected peers. In SIP, that protection is applied to signaling transport. It does not automatically encrypt the separate RTP media stream.
RFC 5630 clarifies the SIPS URI and the use of TLS across SIP hops. It also warns against treating SIPS as proof of end-to-end security. Proxies and other intermediaries participate in routing, and certificate validation and routing policy matter. Verify every hop rather than assuming that a TLS label on the first connection protects the complete path.
DTLS-SRTP is one media-key method
RFC 5764 defines a DTLS extension for establishing keys for SRTP. It is common in architectures that negotiate protected real-time media through DTLS-SRTP. Other platforms may use different supported methods. The decision is not simply whether a familiar acronym appears in a proposal. Confirm how peers authenticate, how keys are derived and scoped, which media legs use the method, and what happens during renegotiation, failover, transfer, conference, or gateway traversal.
Find every place encryption terminates
A call can be encrypted on one leg and clear or differently protected on another. SBCs, hosted PBX platforms, contact centers, PSTN gateways, recording services, transcription tools, voicemail systems, and conference bridges may terminate media or signaling protection because they must route, mix, record, transcribe, or interoperate. That is not automatically a defect, but it is a security boundary that needs an owner and evidence.
Use “end-to-end encrypted” only when the architecture and product documentation show that the communicating endpoints retain the necessary keys and intermediaries cannot access clear media. If an SBC or provider terminates and re-encrypts the call, describe it as protected by segments or legs. The separate VoIP versus landline guide covers other service differences, while the PBX guide explains call-control and carrier boundaries.
Supporting controls still matter
Encryption cannot compensate for a stolen administrator session, a compromised soft client, exposed provisioning credentials, an unsupported phone, or an unprotected recording. NIST SP 800-58 addresses VoIP architecture, availability, physical controls, network separation, perimeter controls, and operational tradeoffs. Current CISA communications-infrastructure hardening guidance reinforces strong cryptography, hardened management, visibility, and removal of unnecessary exposure.
- Require MFA for provider and PBX administration, then limit roles to the minimum required duties.
- Restrict phone, SBC, PBX, switch, firewall, and certificate-management interfaces to approved administrative paths.
- Remove default credentials and unused services. Keep supported firmware and software on an owned update schedule.
- Protect configuration backups, provisioning files, certificates, private keys, API tokens, voicemail, recordings, and exported logs.
- Separate voice devices where that supports management and containment, but remember that a VLAN does not encrypt media.
- Use firewall and SBC policy to admit expected traffic and reduce exposure, but do not describe packet filtering as voice encryption.
- Monitor registrations, administrative changes, certificate failures, unusual call paths, repeated negotiation failures, new integrations, and unexpected recording access.
Rivell’s network design service can map voice paths and management boundaries. Its cybersecurity service can help assign identity, endpoint, logging, incident, and recovery controls around the phone platform.
A VPN protects a segment, not necessarily the full call
A VPN can encrypt traffic between a remote device and a VPN gateway. Beyond that gateway, the traffic follows the platform’s signaling and media design. A VPN does not prove that the provider leg, carrier interconnection, far endpoint, recording service, or stored audio is protected. It can also add routing, latency, MTU, and availability dependencies that must be tested with real call flows.
For remote users, verify the supported client path, home-router and Wi-Fi baseline, device management, split-tunnel behavior, DNS, certificates, emergency-location handling, and fallback. Keep SRTP and TLS controls in scope even when a VPN is present.
Questions to put in the provider proposal
- Which signaling and media protocols are supported on each call leg, and which are enabled by default for our plan?
- Can protected calls fall back to unencrypted signaling or RTP? If so, can policy reject or alert on that downgrade?
- How are SRTP keys established, authenticated, rotated, stored, and removed?
- Which certificates are used, who owns renewal, and how do clients validate peer identity and revocation?
- Where do SBCs, gateways, contact centers, conference services, voicemail, recording, transcription, and CRM integrations decrypt media?
- Which provider employees, subcontractors, customer administrators, and support tools can access configuration or clear audio?
- What logs show protocol negotiation, administrative changes, access to recordings, exports, and failed security controls?
- What happens to signaling, media protection, and keys during transfer, forwarding, failover, disaster recovery, and carrier rerouting?
- How are supported phone and soft-client versions inventoried, patched, revoked, and securely decommissioned?
- Which test evidence and configuration export will be delivered at acceptance and after major changes?
Use Rivell’s VoIP provider selection guide to compare these controls with call flow, support, porting, continuity, and commercial requirements.
Acceptance tests before launch
- Inventory every approved phone, soft client, PBX, SBC, provider, gateway, integration, certificate, and administrative account.
- Place representative internal, inbound, outbound, transferred, forwarded, conference, remote-user, voicemail, and recorded calls.
- Use provider diagnostics and an authorized packet capture at controlled observation points to confirm the expected signaling transport and protected media. Never capture real sensitive conversations for routine testing.
- Verify certificate chain, identity, validity, renewal ownership, failure handling, and alerts at each TLS peer you control.
- Confirm the negotiated SRTP profile, media endpoints, key method, and behavior when the protected option is unavailable.
- Test an expired or untrusted certificate in an approved lab, a disabled account, a revoked phone, an unauthorized administrator, and a policy-blocked insecure call path.
- Verify recording notice and policy, authorized access, retention, export, deletion, encryption, and audit evidence.
- Review logs for the test events and confirm named responders receive actionable alerts.
- Save the approved configuration, network and call-path diagram, certificate inventory, results, exceptions, owners, and retest schedule.
- Repeat bounded tests after carrier, provider, SBC, firmware, certificate, recording, routing, or integration changes.
Caller-ID authentication is a different control
STIR/SHAKEN addresses caller-ID authentication in IP voice networks. The FCC’s caller-ID authentication order focuses on providers signing calls and maintaining the trust framework. That helps evaluate the asserted calling identity, but it does not encrypt the voice conversation. Keep caller-ID authentication and media confidentiality as separate requirements.
Responding to suspected VoIP interception
Preserve relevant logs, call records, configuration history, certificate events, provider notices, and authorized packet evidence. Disable or restrict suspected accounts and endpoints, rotate exposed credentials or keys through an approved process, review new forwarding or recording rules, and contact the provider’s security team. Determine which call paths, users, dates, recordings, and integrations were exposed before making conclusions.
The NIST Cybersecurity Framework 2.0 organizes risk work across Govern, Identify, Protect, Detect, Respond, and Recover. Use that lifecycle to assign phone-platform ownership, evidence retention, notification decisions, corrective actions, recovery checks, and lessons learned.
VoIP eavesdropping FAQ
What is the main technology used to protect VoIP calls?
SRTP is the primary standard for protecting RTP voice media. TLS protects SIP signaling on configured connections. A complete design needs both functions plus authenticated key management and secured endpoints.
Is a VPN enough to stop VoIP eavesdropping?
No. A VPN protects traffic only across its tunnel. The provider, carrier, gateway, recording, integration, and far-end paths still need their own verified protections.
Do firewalls and voice VLANs encrypt calls?
No. They can restrict traffic and improve containment or operations, but they do not encrypt RTP media. Verify SRTP or the documented media-protection method separately.
Does SRTP mean a call is end-to-end encrypted?
Not by itself. SRTP can protect individual media legs. Determine whether intermediaries terminate the protection and can access clear audio before using an end-to-end claim.
How can a business verify that VoIP encryption is active?
Review the saved configuration, negotiated protocol evidence, certificates, provider documentation, authorized packet observations, logs, and negative tests across representative call paths. Retest after material changes.
Plan and test the actual call path
Rivell can document your phones, applications, network, SBC, provider, carrier, recording, and administrative boundaries, then connect the required protections to deployment and acceptance evidence. Request a VoIP security and call-path review or call (856) 603-0000.