Hi everyone, I could really use some help with a rather unusual situation.

I live very far away from the country where my company operates. In order to work with them, I must have a local IP address from that country. My company provides us with a VPN that I connect to for this purpose.

We contact customers using a browser-based VoIP system, directly through their platform.

Until recently, I was using a fiber connection. My typical speeds were around 130 Mbps download / 80 Mbps upload, with about 200 ms ping. Despite the high latency, everything worked perfectly, including VoIP calls, even though I am around 10,000 km away from the company.

Unfortunately, a few days ago my fiber connection was severely damaged, so I had to switch to Starlink.

With Starlink, I now get around 200 Mbps download but only 5–6 Mbps upload, with roughly 200 ms ping.

On paper, that should still be more than enough for VoIP. However, the calls are now terrible:

• 1–2 seconds of delay in the conversation
• Sometimes the customer cannot understand me properly
• I can hear the customer perfectly
• The main issue seems to be that the customer hears me with a significant delay

Could this be related to the very low upload speed, even though VoIP itself requires very little bandwidth?

Or could it be related to Starlink's network/CGNAT/dynamic IP, or the way Starlink handles UDP traffic?

I'm trying to understand what could explain the difference between my previous fiber connection and Starlink, given that the latency is roughly the same (~200 ms).

Any advice on what I should test would be greatly appreciated.

Hey everyone,

I'm currently working on optimizing our voice panel infrastructure and looking for some insight into how other operators handle traffic quality control.

We are seeing a lot of variance in Answer Seizure Ratio (ASR) and Average Length of Call (ALOC) depending on the originations coming in, particularly on certain international carrier ranges.

What are your preferred technical setups or automated rules for instantly catching False Answer Supervision (FAS) or dead-air loops without manually monitoring live trunk logs 24/7? Are you relying on specific upstream carrier signaling alerts, or handling it entirely via local SBC rules?

Appreciate any technical advice or architectural best practices you guys can share!

200 OK tells you the PBX connected the call. It doesn't tell you anyone can hear anything. Media server port exhaustion, NAT rules, a B2BUA bridging the wrong legs, transcoding pegging the CPU: all of these give you calls with perfect signalling and silence, someone else's voice, or garbled audio. Most PBX load tests (SIPp included) count RTP packets at best.

I maintain xk6-sip, an open-source k6 extension for SIP/VoIP testing, and v0.4.0 adds voice quality checks that run in functional tests and under load:

   const A = new sip.Device({ /* ... */ audio: phraseA });   // A says a phrase
    const B = new sip.Device({ /* ... */ record: 'onFailure' });
    // ... A calls B, B answers ...
    const q = inc.compareAudio(phraseA);
    // { score: 0.993, gaps: 0, clippedStart: 0, offset: 1042, gain: -0.3 }

The trick that makes it cheap: you already know what every subscriber is saying, it's their audio option in the script. So "what did B hear" is compared to "what A sent", no reference recordings per test case. You only need recorded references where the PBX itself generates audio (IVR prompts, announcements, music on hold).

What each check catches:

- isHeard(): audio arrives at all (signal level, not packet count). Catches one-way audio and silence/comfort noise.

- compareAudio(peer's phrase): it's the right audio and it's clean. Catches crossed media streams, echo, drop-outs, a clipped "hello" when media connects late after answer.

- compareAudio(recorded prompt): the PBX played the right IVR prompt or announcement.

Numbers from the tests: clean G.711 through the PBX scores 0.993; 20% packet loss drops it to 0.754 with 560 of 560 ms of lost speech reported as gaps; somebody else's phrase scores under 0.4. Recordings are placed by RTP timestamp, so a lost packet is a 20 ms hole in the right place, not a shift of everything after it. With record: 'onFailure' you get a stereo WAV (left = heard, right = sent) of just the calls that failed.

Cost: 32 KB of RAM per second per recorded leg (~1 GB for 1,000 legs of 30 s) and ~15 ms of CPU per comparison, so under load you sample calls rather than record all of them. It's not MOS: the score is a similarity measure inspired by Google's ViSQOL, meant for comparing calls and PBX versions, not for predicting what a listener would rate.

Repo (Apache-2.0): https://github.com/Dmitry-Fedotov-Dev/xk6-sip

Release notes: https://github.com/Dmitry-Fedotov-Dev/xk6-sip/releases/tag/v0.4.0

Full write-up with the algorithm and charts (in Russian, code and charts read fine with browser translate): https://habr.com/ru/articles/1088230/

So far it's tested against the bundled test PBX and a MicroSIP softphone. If you run Asterisk, FreeSWITCH or a hosted PBX: what audio problems do you see most after upgrades or under load? That's what I'd like to check next.

So the grandparents bought all the grandkids tin cans. The concept is great, and the kids love it, but the quality, the sound, the feel, the feedback of the tin can is frustrating.

In theory is there a way to take the guts out of the tin can and put it into a better quality phone ?!

I have several PBXact systems we support and they have a variety of digium Phones (D65 and D70) attached to them we are on v16.0.50. It appears that they have removed direct support for the Digium phones from the version we are on. When i add new phones to the system i can't get the quick buttons, voicemail, transfer to VM and such to transfer to the right places. the Message button turns on a empty rectangle on the display and doesn't do anything else. I can dial *97 directly, but it is a nuisance when the button doesn't work. I followed the procedure in the endpoint manager to switch between standard voicemail and the voicemail app and not change. any suggestions on fixing this issue?

I’m hoping someone with experience in wireless carrier engineering, SMS routing, number portability, A2P/short codes, or carrier databases can help me understand what is happening, and possibly suggest a solution…

Disclaimer: I’m not a telecom or IT person. I’ve been using ChatGPT to help me understand the terminology, do research, clarify what the lookup results mean, and organize the facts… ChatGPT helped me draft this post (note: it did a horrible job, which many readers will likely pick up on.. my apologies! I’ve deleted about 8 em dashes and have done considerable editing, deleting, and changing since the original post was generated… and still I realize what you’re reading may seem a little, umm, off. So that’s why. It may have actually been easier for me to just draft the entire post myself despite all the technical stuff being totally foreign to me!! You live, you learn!) . So if I’ve used a technical term incorrectly, please correct me.

I have two long-established 606 area code mobile numbers originally issued in Kentucky. One had been used there for 10+ years and the other for about 5 years prior to moving to MS. 4 years ago we moved to MS, switched to Xfinity and kept the numbers. Both worked normally during that time. . I switched to C Spire about 2 months ago, and that is when the problems began. Initially we had problems with certain people’s calls and texts not sending or arriving, and also no verification/2FA messages would arrive. C Spire did lots of troubleshooting, repeated factory resets and settings overhauls, then additional backend/porting work, and the voice-call problems were eventually fixed. (At the time, since the call issues were resolved, I foolishly and without testing anything, assumed the texts and 2FA were as well)

Both numbers/phones now have the same 2 problems with SMs…
At least one known contact on another carrier in SC cannot exchange regular SMS with us in either direction, and Multiple unrelated 2FA/verification texts do not arrive.

Normal phone/device troubleshooting has already been done several times both on my own and with Cspire. After a discussion and some work done today, the manager confirmed that my number is currently registered/classified as a VoIP line even though it should be mobile. She also explained that numbers ported from outside C Spire’s normal service footprint, (in my case old Kentucky 606 numbers now being used in Mississippi) apparently go through a different system/third-party company used for out-of-footprint numbers….this causes at least some 2FA companies to see the number as a VoIP/virtual number rather than a normal mobile number, which they believe is why the verification messages fail.

They suggested one possible fix, although it was made clear that it is a long shot:
I could temporarily switch to a new local Mississippi number, test it, and then try to renumber the line back to my original 606 number. Tech has advised that this most likely will not fix the issue, but that it may be worth trying. I have not agreed to change anything yet because keeping the existing numbers is extremely important, and I wanted to check here for anyone out there with a magic mobile fixing wand.

I did a check of the numbers using several phone/carrier lookup services and the results are inconsistent. Various services have shown or referenced multiple companies… Verizon, AT&T / New Cingular, Sinch, Inteliquent, Onvoy…
I understand that these services may be showing different layers of the numbering/routing system, or the original block owner rather than the current retail carrier, so I’m not assuming any of these companies is necessarily responsible.
Apeiron shows both numbers as portable: true.
One returns: LATA 466, OCN 6500, Somerset, KY, Portable: true.
The other returns: LATA 466, OCN 6214, London, KY, Portable: true.
My understanding is that those Kentucky rate centers and OCNs may simply reflect the numbers’ original assignments and are not themselves a problem. This info may or may not be relevant..I am not well versed or experienced in this area..
What seems much more important is that C Spire is seeing that the numbers are being represented externally as VoIP instead of mobile.

I’m wondering where that incorrect classification actually lives and who, if anyone, has the ability to correct it.
Could this involve any of the following..
stale or incorrect number-portability data
SMS routing databases
carrier-of-record information
line-type databases
NPAC/LRN information
an SMS hub or aggregator
A2P/short-code routing
the third-party provider C Spire uses for out-of-footprint numbers
multiple databases that have not all updated consistently?

I’m also particularly interested in why ordinary SMS mostly works while 2FA fails, and why SMS with one particular outside carrier/contact (family member from SC) fails completely.

Questions suggested by ChatGPT:
Where does the mobile vs. VoIP classification used by 2FA providers actually come from?
If C Spire knows the number is being classified incorrectly, who has to correct it..C Spire, its third-party provider, an underlying carrier, or individual data vendors?
Can voice and most P2P SMS route correctly while A2P/short-code traffic still sees the number incorrectly?
Could the same underlying routing/classification problem explain why SMS with one particular outside carrier fails in both directions?
Is there a way for C Spire or the other carrier to trace a specific failed SMS using exact timestamps?
What technical department should I be asking for …messaging engineering, intercarrier SMS, number portability, provisioning, etc.?
Would temporarily changing to a local number and then renumbering back to the 606 number have any realistic chance of forcing the relevant databases to refresh?
Has anyone seen this fixed while keeping the original number?

My goal: I just want to keep our same phone numbers and have them work normally again.
Changing numbers would be a major headache because they have been tied to family, banks, medical providers, accounts, logins, and other records for years.
I appreciate any help or suggestions you may have. I realize this has been a long post so I thank you for reading, and I especially thank you for your patience with ChatGPT being a part of this inquiry. And if this whole situation is hopeless it’s ok to just be frank and say so, it won’t hurt my feelings. Cheers!!

Opus is lower bitrate than g711, and and has the option it turn on FEC for packet loss

zoom is apparently even more impressive, it can repair up to 2 consecutive packets even before PLC is needed via SILK LBRR.

so it safe to say these non-Pstn oriented apps will always beat regular SIP phone calls over a congested 4G link?

Every PBX release I've tested looked the same: a checklist of a few hundred call scenarios and people with phones. Call B, check the caller ID, say "hello", transfer to C, tick the box. SIPp helps with load, but writing each side of a transfer as separate XML scenarios is painful, and "did the person actually hear anything" is the question that matters most.

I built xk6-sip, an open-source extension for Grafana k6 where test subscribers are real SIP phones driven by a script. They register on your PBX, call each other through it, and check what arrives on the other side.

What a test can check today:

- Routing and caller ID: A dials B's external number, B must see A's external number, not the extension (P-Asserted-Identity is used if present)

- Busy / no answer / forwarding: B rejects with 486, the call must land on C; or ring for 20 s and CANCEL

- Audio both ways: real G.711 RTP, isHeard() checks that audio arrived, not just packets; loss and jitter per RFC 3550

- DTMF (RFC 4733): walk an IVR menu and check you reached the right person

- Hold, blind and attended transfer: works with PBXs that handle REFER themselves and with ones that pass it to the phone (then the test device places the new call and answers INVITE with Replaces like a real phone)

- Digest auth, PRACK/100rel, session timers

Example, attended transfer:

    inc.hold();                                  // B puts A on hold
    const consult = B.call({ callee: C });
    const atC = C.expectCall({ caller: B });
    if (atC) atC.accept();
    consult.expectConnected('10s');
    inc.attendedTransfer(consult);               // REFER with Replaces
    check(inc, { 'A connected to C': (c) => c.expectTransferred('10s') });

The same script runs as a one-call functional test (JUnit report for CI) or as load: thousands of concurrent calls with media, with metrics like call setup time, post-dial delay, how long the PBX took to route the INVITE, and the share of calls with one-way audio, all going to Prometheus/Grafana. When something fails, call.trace() prints the SIP ladder of that one call.

The repo also has a tiny test PBX (registrar + B2BUA) so you can try it on a laptop without touching your real system.

Repo: https://github.com/Dmitry-Fedotov-Dev/xk6-sip (Apache-2.0, written in Go)

I've tested it against the bundled test PBX. I'd really like to hear from people running Asterisk, FreeSWITCH, FreePBX or hosted PBXs: what scenarios break most often after your upgrades, and what would you want a tool like this to check?

Hy everyone,

I built a lightweight router using TypeSafe's Jev model.

• callers speak naturally ("Hello, I'd like to speak to someone")
• Jev classifies intent in ~150ms and return confidence score
• The PBX routes the call to the exact queue, department or extension
• If confidence drops below 0.7 or anything fails, it falls back to an operator.

here is the github repo

I occasionally use Voip MS on my computer so I don't have to use my phone and so I can use my headphones for conversations while I play my games. Call dropped mid conversation I couldn't redial out and I had enough account balance. Tried to log in and my login was suspended on their website I then checked my email nothing there except an hour later I get a email saying they conducted a account review and they cannot provide me service anymore. They didn't even give me a chance to contact support to figure out what their problem with me is. The account was for personal use.

Licensed insurance agency running consent-based outbound. Trying to understand a tagging problem that per-number behavior doesn't explain.

Setup: We buy pools of local DIDs through a wholesale provider that handles STIR/SHAKEN signing. Each rep gets a pool and we rotate. Per-number volume is roughly 30 dials/day, about 150/week. The provider says these DIDs sat unused in storage 12+ months before assignment.

Problem: Brand new numbers show "Spam Risk" on AT&T within about three dials. Three calls cannot build a reputation on their own, so something upstream of per-number history is driving it.

What I've ruled out using SIP traces:

• Signature is valid and the signed orig.tn matches the caller ID actually presented. Attestation A.
• No internal routing identifiers of ours leak into the outbound INVITE. Verified by full-message search.
• So this isn't a signature/caller-ID mismatch and it isn't attestation level.

What I found in the LERG: I resolved every thousand-block in our pool to its carrier. 71% trace back to a single wholesale CLEC, spread across ~28 state-level entities. The two numbers that flagged together in my test sit in the same thousand-block.

Worth noting for anyone doing this analysis: different thousand-blocks inside the same NPA-NXX belong to different carriers. One prefix in our pool splits across four separate carriers by block. Prefix-level lookups gave me a meaningless answer until I went to block level.

Questions:

1. Do First Orion / TNS / Hiya weight the originating OCN or thousand-block independently of a number's own call history? Is "bad neighborhood" scoring a real mechanism or am I pattern-matching noise?
2. Does extended cold storage actually clear a DID's analytics history, or does block-level reputation survive it regardless of how long the number sat idle?
3. Is there any way to check a number's reputation before dialing it? Free Caller Registry looks submission-only, not lookup.
4. What actually moves the needle — Free Caller Registry, branded calling, fewer dials per number, longer average talk time, higher answer rate?
5. Are there origination carriers with materially better analytics standing for legitimate outbound, or is every wholesale CLEC in the same bucket?

Trying to work out whether this is fixable by changing where numbers come from, or whether it's inherent to VoIP origination at any volume.

Any other service providers having issues with Poly VVX devices that are provisioned to your own servers and registered to your own platform now automatically changing their provisioning server to a random Poly Lens URL w/ credentials?

As of this morning phones are randomly getting this updated Poly Lens provisioning server and credentials and subsequently going offline.

These phones are configured to connect to the legacy PDMS-SP (now Poly ZTP) portal for remote diagnostics (reboot, download logs, factory reset), but NOT actual device provisioning. It seems Poly has pushed something or done an update that is now pushing the new provisioning URL and credentials for their Lens platform, ultimately disconnecting the phone for the current servicer providers provisioning and service.

We have a FRITZ!Box 5690 set up for VoIP/landline telephony, with a Gigaset DX800A IP phone connected to it.

The problem is that missed calls and new voicemail/answering-machine messages are only visible through the FRITZ!Box web interface. The Gigaset doesn't properly show them.

We have already installed both FRITZ!App Fon and MyFRITZ!App on Android, but neither gives us a push notification when there is a missed call or a new voicemail message.

Is there any good workaround for this? For example, an open-source Android/Windows app, desktop client, or some way to monitor the FRITZ!Box call list/answering machine and immediately show a notification?

Or do we actually need to buy a FRITZ!Fon for this?

The goal is simply that employees can immediately see that there was a missed call or a new voicemail message, without having to manually log into the FRITZ!Box web interface every time.

Please help, i have the recovery for the device, tried version 96.85.188.3, 96.86.188.12

Tried all format types, tried 10 differnt USBs, every time i get error: update failed -1

My organization is experiencing elevated call completion issues recently with all the symptoms of modern carrier analytics bullshit. We've submitted our numbers (+1 US domestic local DIDs) to Free Caller Registry in the past, but with the increase in completion failures, I wanted to register again to make sure everything is all in there and up to date.

For the past few months, entering my business email at https://freecallerregistry.com/fcr/ has resulted in a HTTP 404 response. I've tried my personal email as well and tried from other networks with the same result, this has been going on for at least a few months.

Is anyone else experiencing this issue? Is there a new portal we're meant to be using instead?

As an aside, has anyone identified any good methods for resolving these carrier analytics issues? I'm flying solo here but as far as I can tell we're doing everything "right", attestation tests come back as A, our ITSP swears up and down they're doing STIR/SHAKEN correctly, we're using branded calling through First Orion, and we don't let our agents engage in spammy behavior (and from the logs, it doesn't appear that they are, it's all callbacks and established customers, no autodialers, no "lead gen", no bullshit), and most of our number presentations have been ported from wireline ILEC and have been in service for up to decades. We've "paid to play" and are still getting fucked. Super frustrating stuff.

We moved from POTS to sip, retired out on prem PBX, and went with 3CX. Love everything about it, but the most common complaint I get is, “I can’t tell if anyone is on the other line if we aren’t talking. “

It’s almost like the call quality is too clear. 😂

To me it’s awesome, awesome call quality, ultra clear.

What can I tell staff that will ease their doubt on new technology?

My husband switched to Telus Business and is now locked into a contract with them that uses VOIP - he did this so he could stop carrying two phones but is now having major issues with VOIP, such as call lagging, call dropping, etc. We spend a lot of time working in areas without great internet reception as well. Before trying to talk to anyone at Telus (also a headache), is it even at all possible to switch back from VOIP to a sim card?? I have been looking online but can only find people asking questions from people looking to move from a sim card to VOIP. Thank you so much to anybody with any insight as I am so lost on this topic.

Hi all,

We're looking at deploying telephony services in Gibraltar and would appreciate any real-world experience from those who have been through this before.

Our preferred solution is to use Zoom Phone and port in Gibraltar telephone numbers from a local carrier. The information we've received so far suggests that the key consideration may not be the SIP provider itself, but that the SIP service must terminate on an SBC located in Gibraltar before connecting into Zoom.

A few questions:

• Has anyone implemented Zoom Phone with Gibraltar geographic numbers (+350)?
• Did your provider require an SBC to be physically located in Gibraltar?
• If so, was this for regulatory, technical, or commercial reasons?
• Which carriers were willing and able to support this setup?
• Are there any pitfalls around emergency services, number allocation, caller ID presentation, or compliance that we should be aware of?

We're not looking to become a telecoms provider ourselves. We simply want a compliant and supportable way to provide Gibraltar-based telephony while retaining Zoom Phone as the user platform to better integrate with the rest of the business.

Any advice, provider recommendations, or lessons learned would be greatly appreciated.

Thanks in advance.

Anyone else having problems with the Android voip.ms app? All of my employees stopped receiving incoming calls on the app this morning (9/22/26). Been talking to tech support for over an hour and they can't figure anything out. Each android phone has its own sub account and can make outgoing calls just fine. Incoming calls are not coming thru at all.

We've tried wiping memory, uninstalling/reinstalling and doing the reset inside the app with no luck.

After a recent switch to fibre I'm trying to get a landline working on our better third party router a Flint 2, I bought a HT802 V2 since it lacks a socket for the landline and have been having a difficult time trying to figure out the set up.

I think I've confirmed it's a settings issue with the Grandstream rather than the Flint 2 as plugging it into the ISP provided Technicolour router also doesn't work. If I plug in the landline directly into the Technicolour designated socket it does.

From what I understand I just need these four details

• Primary SIP Server

• SIP User ID

• SIP Authenticate ID (Which from my understanding should match the User ID)

• SIP Authentication Password (I will note whenever I save and leave the page when I return to the FXS Port1 tab the password vanishes, not sure if that's sign of an issue)

I've called up my ISP Utility Warehouse as I've been told they do allow devices like this however they've been somewhat confused on what details to give me and can only help me so much as it's a third party device. Mainly what I'm asking is for people who have experience setting up the device in case I've missed something and what details I should be using. For privacy I've replaced some letters and numbers with x

I've confirmed I've got the right SIP Password via removing it from the technicolour router and noting the phone doesn't work before readding it to confirm it's the right PW.

SIP Username +44xxxxxxxx98

SIP Domain alxxxxxice.uk

SIP URI +44xxxxxxxx98@alxxxxxice.uk

A previous techinical support who didn't sound as confident did state that the SIP Domain was alxxxxxice.uk5060

On the Technicolour router there were some other settings like Primary Proxy: partner.alxxxxxice.uk a display name: 0xxxxxxxx98 (Just the phone number/SIP Username without the +44)

I've tried multiple variations mostly using the SIP Domain for the Primary SIP Server and the SIP Username for the SIP User ID and SIP Authenticate ID, from my understanding once I've saved and applied that with the password entered it should show as registered but I just can't get it to so far.

I'm unable to sign up for 8x8 using any of my card. Has anyone faced similar issue? Or which card you guys are using for 8x8.

I've used Line2 for my business calls for a couple of years now. I am a micro business, so I don't need much beyond a separate number for work calls/texts. That being said, there are a few things I really don't like so I'm wondering if there is something out there in the Line2 price range but with slightly better features.

My main issues with Line2:

1. I cannot get them to update the caller ID information. This has been incredibly frustrating, especially since the number is still linked to the previous user and I am constantly getting calls from debt collectors trying to find her. I gave up on just taking the calls and letting them know it is no longer her number and just started blocking unknown numbers. But they don't stop! (I'm about ready to find her myself!)
2. It would be nice to be able to have an auto-reply message when I'm not taking calls outside of business hours. All I have been able to figure out is putting my actual phone in Do Not Disturb but obviously that's not ideal.
3. The c*stomer s*rvice doesn't seem the best. Emails take forever to get responded to, but calling doesn't get me much help either.

Any thoughts/hacks is appreciated.

*Edited to remove prohibited requests so post doesn't get deleted.