Connection state
Confirm the intended publisher and receiver are connected, then compare the address, port, Stream ID and passphrase before changing media settings.
SRT ingest, routing and operations
An SRT server receives and forwards live video over Secure Reliable Transport. It gives encoders and receivers a reachable endpoint, so a camera feed can travel from a remote location to a production system.
Callaba brings SRT connections, Multiview monitoring and main/backup routing into one browser dashboard. Receive a finished programme, or use your incoming cameras in the Video Mixer. Run the software in your AWS account or on infrastructure you control.
Transport model
SRT carries media over UDP and can request retransmission inside the latency window. AES encryption is optional. Caller initiates a connection and listener waits; neither defines video direction. Rendezvous is the protocol mode where both peers initiate toward each other. The connection examples below use caller and listener.
| Job | Connection |
|---|---|
| Encoder sends to Callaba | Encoder: caller → Callaba: listener |
| Callaba pulls from a remote source | Callaba: caller → source: listener |
| Receiver pulls from Callaba | Receiver: caller → Callaba: listener |
| Callaba pushes to a remote receiver | Callaba: caller → receiver: listener |
Live SRT demo
Create a temporary connection for one hour. Send your feed to the publisher address in Caller mode, leaving Stream ID empty. Watch it in Multiview below, or connect your receiver to the receiver address in Caller mode.
Operate the endpoint
Confirm the intended publisher and receiver are connected, then compare the address, port, Stream ID and passphrase before changing media settings.
Incoming bitrate proves arrival, not decode. RTT is round trip time, not end-to-end latency. Compare loss and drops over the same interval.
Open server logs for connection events. Search and pause the view, then export the relevant period when a source or network event needs correlation.
For a first test, MPEG-TS carrying H.264 video and AAC audio is a practical starting point. An SRT relay is different from decoding or browser preview: it moves the incoming encoded media. Check the receiver’s container, video codec and audio-track expectations before treating a connected transport as a working programme.
Own SRT server
Choose how your encoder and receiver connect. Replace the placeholders with the values shown in your Callaba dashboard.
Host & port patternTwo distinct UDP ports: one publisher port and one receiver port. Stream IDs are deliberately empty.
srt://example.invalid:YOUR_PUBLISHER_PORT?mode=callerUse the publisher UDP port. Leave Stream ID empty for this separate-port pattern.
srt://example.invalid:YOUR_RECEIVER_PORT?mode=callerUse the distinct receiver UDP port. Leave Stream ID empty here too.
In Streaming servers → SRT servers, create the listener and reserve the UDP port or ports. In Streams, permit an empty Stream ID for the intended publisher and receiver roles.
Set the encoder and receiver to Caller and paste their respective address. Caller and Listener describe who initiates the session; they do not describe media direction.
Start the source. Confirm the publisher and incoming bitrate in Statistics, then check the picture and listen to the audio at your receiver.
After ingest
Add the modules your production needs. For browser viewing, use Multiview or a web-player delivery path: a browser cannot play a raw SRT URL.
Check incoming sources side by side before routing a source or building a programme.
Open MultiviewRecord an incoming source, or keep a separate recording of the finished programme after a mixer.
Open RecordingUse a delivery task with the destination URL and stream key when the destination expects RTMP.
Open RTMP deliveryBuild scenes from input sources, prepare Preview, then take Program with CUT or FADE.
Open Video MixerFor SRT-to-NDI delivery, check that the receiving NDI environment can discover and receive the published source. A remote SRT connection does not extend local NDI discovery across the internet.
Explore the NDI workflow →See the workspaces: Open Multiview demo ↗Open Mixer demo ↗
Starting with an empty, unfiltered SRT servers list? Choose Create From Template → SRT Main/Backup Failover → Create this preset to create source servers, a routing output and a Multiview board. Keep working servers on an existing installation.
Route selection is separate from CUT/FADE scene transitions. Do not assume a switch is interruption-free; measure any freeze or silence in your actual output.
Read the failover guide →Where the server runs
Callaba provides the browser workspace. Your team operates the host, network access, and storage. Check the installation requirements and available modules for your chosen release before deploying.
Review current software terms in AWS Marketplace. EC2, storage, and data transfer remain infrastructure costs.
Deploy on AWS ↗Follow the current self-hosted Linux installation guide for host requirements and setup instructions.
Install self-hosted →Count simultaneous sources and outgoing copies. Leave network headroom for retransmissions, and account for CPU when recording, Multiview, mixing, or transcoding join the path. Test the intended workload under load before choosing an instance.
Connection checks
Verify the enabled server, public address, publisher UDP port, firewall or security group, then the permitted Stream ID and passphrase on both ends.
Check bitrate first, then the receiver selection and its supported container, video codec, and audio tracks. Transport arrival alone does not prove decoding.
Compare bitrate, RTT, packet loss, and drops over one interval. Check uplink capacity and retransmission headroom, then correlate disconnects against exported logs.
Deploy Callaba on AWS or a supported Linux host, then create an enabled SRT server. Set the publisher and receiver ports, latency and permitted connections in Streams. Allow the assigned UDP ports through the firewall. Connect an encoder to the publisher address and a decoder or production application to the receiver address, then check picture and audio at the receiver. Follow the connection setup above for separate ports or paired Stream IDs.
SRT is an open-source protocol, and open-source servers such as MediaMTX can receive and forward it without a software licence fee. You still provide the host, network capacity and maintenance. Callaba is paid software with browser management and optional production modules. Its calculator on this page shows self-hosted and AWS software costs; infrastructure, storage and traffic are separate. The one-hour demo lets you test a connection before deploying your own server.
Allow UDP on the listener ports assigned to your SRT server; opening a TCP port with the same number is not enough. In Callaba’s separate-port setup, the encoder uses the publisher port and the receiver uses the receiver port. With paired Stream IDs, both connect to the configured shared listener port. Behind a router, forward the required UDP ports to the server. On a cloud host, check both its security group and operating-system firewall, restricting source addresses where practical.
For the Callaba demo and the connection examples on this page, set both the encoder and receiver to Caller. Callaba listens on the assigned ports. Caller and Listener describe who starts the connection, not who sends video: a Caller can publish or receive. Rendezvous is another SRT connection mode, but it is not used in these Callaba examples and is not a substitute for checking firewall reachability.
No. A Stream ID selects a stream or access role; a passphrase is used for SRT encryption. This demo accepts empty Stream IDs on separate publisher and receiver ports. On your own Callaba server, the allowed connections in Streams must match that setup. For a shared-port connection, use the paired publisher and receiver IDs from the dashboard. If encryption is enabled, configure the matching passphrase at both ends. See the two connection methods.
Choose it from the measured network RTT and packet loss, not a universal low-latency preset. The demo starts at 200 ms; a long-distance or unstable link may need more recovery time. Raise the buffer if retransmitted packets arrive too late, then retest the actual feed. More buffering cannot fix insufficient bandwidth. Callaba’s field uses milliseconds; FFmpeg’s URL option uses microseconds, so latency=200000 means 200 ms. This buffer is only part of the total camera-to-screen delay. Follow the SRT latency tuning guide.
In OBS, open Settings → Stream, select Custom and paste your Callaba publisher SRT URL into Server. Leave Stream Key empty. For this demo, use the generated publisher address in Caller mode without a Stream ID. Start streaming, check the incoming bitrate in Callaba, then verify the feed through Multiview or your receiver. If the connection works but the picture does not, check the video and audio formats. See the OBS SRT sending guide for the full setup.
To send, configure a vMix SRT output as Caller using Callaba’s publisher host and port. To receive, choose Add Input → Stream / SRT → SRT (Caller) and enter Callaba’s receiver host and port. Set latency in milliseconds, and supply a Stream ID or passphrase only when your server configuration requires it. The publisher and receiver ports are different in this page’s demo. Follow the vMix sending and receiving guide.
A browser cannot open an srt:// address directly. Callaba receives the SRT stream and provides a separate browser preview through Multiview. The demo on this page lets you check your incoming feed that way. For audience playback, configure a browser delivery output and test its picture, audio and delay separately; an operator preview is not the same as an audience player. The SRT Player page covers the receiving options.
Yes. Receive the SRT feed in Callaba, then configure a Re-streaming task with that source and the destination’s ingest URL and stream key. An SRT URL is not an RTMP ingest address. Check that the destination accepts your video and audio codecs; if it does not, add transcoding rather than just forwarding the encoded feed. Verify playback at the destination. See RTMP delivery options.
SRT transports an encoded feed; it does not itself mix scenes or change codecs. Callaba provides those as separate processing functions. Add SRT sources to the AV Mixer, arrange the next scene in Preview and take it to Program with CUT or FADE. Each camera or other mixer input counts as one unit; other active modules are counted separately. See the Video Mixer and the pricing calculator.
From an empty, unfiltered SRT servers list, choose Create From Template → SRT Main/Backup Failover → Create this preset. On an existing setup, use SRT Routing → SRT main/backup failover with Pull routing and add the source hosts. In an authenticated Multiview, open the routed output tile’s Open input menu → Source route to select main or backup; public viewers cannot switch it. Rehearse a main-source interruption and verify video and audio at the receiver. Do not assume the switch is interruption-free.
A managed service such as Castr operates the infrastructure for you. With self-hosted software such as Nimble Streamer, MediaMTX or Callaba, you operate the host. Callaba adds browser controls, Multiview, recording and mixing to the SRT connection. Compare who will maintain the server, where it needs to run, and what happens to the feed after ingest. For Callaba, software licensing and infrastructure remain separate costs. If routing between endpoints is your main job, see the SRT Gateway workflow.
API access
This read-only request returns configured SRT servers. Replace the host and token with your server’s values; it does not alter configuration.
curl --request POST 'https://YOUR_CALLABA_SERVER/api/srt-servers/getAll' \
--header 'x-access-token: YOUR_API_TOKEN' \
--header 'Content-Type: application/json' \
--data '{"limit":10,"skip":0,"sort":{"server_created":1}}'FFmpeg publisher and receiver commands, latency units and a repeatable interruption test for a Callaba SRT server.
An SRT connection has an initiator and a listening endpoint. The sender and receiver roles determine where the media goes. In the separate-port test below, both FFmpeg applications are Callers: the publisher sends video to Callaba, while the receiver requests it from Callaba.
SRT also defines Rendezvous, in which both peers attempt to establish the connection. It requires compatible peer configuration and a suitable network path. The examples here use Caller and Listener; they are not a Callaba Rendezvous configuration.
Protocol details: Haivision SRT socket options.
Install an FFmpeg build with SRT support. Create a Callaba SRT server, allow publisher and receiver access in Streams, and copy its host and assigned UDP ports. These examples use separate ports and empty Stream IDs. A shared-port setup needs the paired Stream IDs shown in the connection guide above.
Allow the chosen UDP ports in the host firewall and, where applicable, the cloud security group or router. A private address needs a reachable path from the sender. Leave any public firewall rule as narrow as the source and receiver networks allow.
Use a test file containing video and audio that the receiver can decode. The command copies the existing encoded tracks into MPEG-TS; it does not convert an incompatible codec. For an encrypted connection, configure matching SRT passphrases at both ends rather than publishing a credential in this example.
Replace YOUR_CALLABA_HOST, both assigned ports and input.ts. The values below are placeholders, not a public test server. Start the publisher first.
ffmpeg -re -stream_loop -1 -i input.ts -map 0:v:0 -map 0:a:0 -c copy -f mpegts "srt://YOUR_CALLABA_HOST:YOUR_ASSIGNED_PUBLISHER_PORT?mode=caller&latency=200000"
In Callaba Statistics, the publisher should appear with a nonzero incoming bitrate. If it does not, check connectivity and access before changing the media settings.
ffplay "srt://YOUR_CALLABA_HOST:YOUR_ASSIGNED_RECEIVER_PORT?mode=caller&latency=200000"
Confirm moving video and audible programme sound. A connected receiver with a blank picture can still have a media-format or stream-selection problem. Stop each command with Ctrl + C.
FFmpeg uses microseconds for this URL option: latency=200000 means 200 ms. The Callaba latency field uses milliseconds. This is a starting value for a test, not a recommendation for every network. See FFmpeg’s SRT options.
The SRT recovery buffer gives retransmitted packets time to arrive. If packets arrive too late, raising the buffer may help at the cost of extra delay; it cannot create missing bandwidth. Observe RTT, loss and late drops while the actual encoder is running, then change one setting at a time.
Glass-to-glass delay also includes capture, encoding, processing and playback. Measure that full path when an operator needs to talk to a remote contributor. Use the SRT latency guide for the tuning procedure and the statistics guide to interpret the counters.
Run this check on a test feed or during a planned maintenance window. Keep the real receiver visible and audible throughout.
Keep the configuration and observations with the event setup. The result applies to the feeds, receiver and network tested, not to every SRT deployment.
An SRT receiver may be a decoder, production application or another server. A browser needs a separate playback path. Callaba Multiview provides an operator preview; an audience player needs its own configured delivery output. Use the SRT Player guide when the task is receiving and watching a feed.
Dashboard, media management, internal disk and connected storage controls, domains and SSL, cluster management, and supported configuration migration are part of every Callaba deployment. Add functional units only where your live workflow needs capacity.
On-premises and self-hosted use the same modular license. Start with the functional capacity your workflow needs.
AWS Marketplace uses hourly software pricing. Choose the smallest C5 instance that covers the busiest part of the workflow.
* The trial applies to Callaba software. AWS infrastructure is charged separately. See AWS Marketplace for current trial and subscription terms.
Open the live Multiview demo, launch Callaba on AWS, or install it on your own Linux server.