A VoIP proposal for a medical office reads like a different language: SIP trunks, seats, codecs, SBCs, QoS, 10DLC. Vendors rarely define the terms because they use them all day, and buyers rarely ask because nobody wants to look uninformed. This glossary covers the vocabulary that shows up in quotes and demos, grouped by what it affects, with a note on why each matters when the phones are the front door of a clinic.
Architecture terms
VoIP (Voice over Internet Protocol). Phone calls carried as data packets over an IP network instead of dedicated copper circuits. Everything else here sits on top of this idea.
PBX (Private Branch Exchange). The system that routes calls inside an organization: extensions, auto-attendants, hold, transfer, voicemail. A hosted PBX (also called cloud PBX or UCaaS) runs in the vendor's data center and you rent it per user. An on-premise PBX is hardware or software you own and maintain in your building.
SIP (Session Initiation Protocol). The standard signaling protocol that sets up, modifies, and ends calls. Nearly all modern VoIP is SIP-based. A SIP trunk is a virtual connection between your PBX and a carrier that carries a set number of concurrent calls, and is the usual way an on-premise PBX reaches the outside world.
Seat, user, or extension. The unit most hosted services bill by. A seat is typically one person or device with a login. Shared devices such as a waiting-room phone may be priced differently, so ask.
Softphone. An application on a computer or mobile device that acts as a phone. Relevant for remote staff, after-hours triage, and anyone who moves between rooms.
ATA (Analog Telephone Adapter). A small device that lets an analog phone or fax machine connect to a VoIP service. Fax over VoIP is unreliable without special handling; see T.38 below.
Call quality terms
Codec. The algorithm that compresses voice into data. G.711 uses more bandwidth and sounds like a traditional landline; G.729 uses far less bandwidth at some cost in fidelity; Opus and G.722 are wideband codecs that sound better than a landline on a good connection. Which codec a vendor uses determines how much bandwidth each call consumes.
Bandwidth per call. Roughly 90 to 100 kbps each direction for G.711 including overhead, and about a third of that for G.729. Multiply by your peak number of simultaneous calls to size your connection.
Latency. The one-way delay between speaking and being heard. Under about 150 milliseconds is comfortable; above that, people start talking over each other.
Jitter. Variation in packet arrival timing. High jitter causes choppy, robotic audio. Phones use a jitter buffer to smooth it out, at the cost of a little latency.
Packet loss. Packets that never arrive. Even one percent loss is audible on a voice call.
QoS (Quality of Service). Network settings that prioritize voice packets over other traffic such as imaging uploads or software updates. QoS on your local network and router is the single biggest lever for call quality and is your responsibility, not the vendor's.
MOS (Mean Opinion Score). A 1-to-5 rating of perceived call quality. Vendors' monitoring dashboards often report it; above 4.0 is good.
T.38. A protocol for sending fax reliably over IP. If you still fax referrals or prescriptions, confirm the provider supports it or offers a hosted e-fax alternative.
Feature and routing terms
Auto-attendant (IVR). The menu that answers and routes callers. Language options matter for practices serving patients with limited English proficiency.
Ring group and call queue. A ring group rings several phones at once or in sequence. A queue holds callers in order, plays hold messaging, and distributes to available staff, with reporting on wait times and abandonment. Front desks with more than two people answering usually need a queue, not a ring group.
Call park, transfer, and warm transfer. Park places a call on a shared hold slot anyone can pick up. A warm (attended) transfer lets staff speak to the recipient before connecting the patient; a cold (blind) transfer does not.
Voicemail-to-email and transcription. Convenient, but it moves voice messages that may contain patient information into email. Confirm how that is secured.
Number porting. Moving your existing phone numbers to the new provider. Ask about timelines and whether the vendor manages the process.
E911 and dispatchable location. Federal rules require multi-line systems to allow direct 911 dialing without a prefix, to notify a central point when 911 is dialed, and to convey a dispatchable location such as a specific suite or floor. Ask how the system handles this for remote workers and softphones.
SMS, A2P, and 10DLC. Application-to-person texting from a business number. Carriers require business numbers used for texting to be registered under the 10DLC framework; unregistered traffic is filtered. Relevant if you plan to text appointment reminders.
Security terms
TLS and SRTP. TLS encrypts SIP signaling (who is calling whom); SRTP encrypts the voice stream itself. Ask for both. A provider that offers only one is leaving half the conversation exposed.
SBC (Session Border Controller). A device or service that sits at the edge of a voice network to police SIP traffic, block attacks, and handle protocol quirks between networks. Hosted providers run them on their side; on-premise deployments should have their own.
Toll fraud. Attackers compromise a phone system and pump expensive international calls through it. Ask about international call blocking, spending limits, and anomaly alerts.
BAA (Business Associate Agreement). If the provider stores voicemails, recordings, transcriptions, or call detail that includes patient information, HIPAA requires a BAA. Ask which specific services the BAA covers, because some vendors exclude texting or transcription.
STIR/SHAKEN. The caller-ID authentication framework carriers use to fight spoofing. A provider that signs your calls with full attestation helps keep your outbound calls from being labeled as spam on patients' phones.
Pricing and contract terms
Per-seat versus per-line. Hosted services usually bill per seat monthly. SIP trunking bills per concurrent call path. Compare like with like.
Metered versus unlimited. Unlimited usually means unlimited domestic calling within a fair-use policy. Read the policy; call centers and auto-dialers often fall outside it.
Regulatory recovery fees and USF. Line items added on top of the quoted rate to recover federal Universal Service Fund contributions and other regulatory costs. They can add ten to twenty percent to the headline price, so ask for a sample invoice.
SLA (Service Level Agreement). The uptime the vendor commits to and what you get if they miss it. Note whether the SLA covers only their platform or also the path to your office.
Term, auto-renewal, and early termination. The length of the commitment, whether it silently renews, and the penalty for leaving early. Also ask who owns your phone numbers at termination.
Putting it together
A well-specified clinic quote should tell you, in these terms: the number of seats and shared devices, the codec and expected bandwidth, whether TLS and SRTP are on by default, how E911 dispatchable location is handled, whether a BAA is available and what it covers, the SLA, the fees beyond the per-seat rate, and the term. If a proposal is silent on any of those, the gap is your next question.
Common questions
What is the difference between a hosted PBX and a SIP trunk?
A hosted PBX is a complete phone system run in the vendor's cloud and rented per user; you only need phones and internet. A SIP trunk is just the connection between a PBX you own and the public phone network, billed by concurrent call paths. Most small practices choose hosted; larger organizations with existing PBX investments often use SIP trunks.
How much internet bandwidth does a VoIP call use?
About 90 to 100 kbps in each direction for the G.711 codec including overhead, and roughly a third of that for compressed codecs like G.729. Size your connection for peak simultaneous calls plus your other traffic, and configure QoS so voice is prioritized.
What do TLS and SRTP mean on a VoIP quote?
TLS encrypts the signaling that sets up calls, and SRTP encrypts the audio itself. A clinic should expect both to be enabled by default, since together they protect who is calling whom and what is said.
What is 10DLC and why does it matter for a medical office?
10DLC is the carrier framework for registering standard ten-digit business numbers that send application-to-person text messages such as appointment reminders. Unregistered messaging is filtered or blocked, so any practice planning to text patients needs its provider to register the campaign.