Turn your broadcast streams into immediate decisions
SondaStream (Sonda Stream) is a professional broadcast probe that centralizes real-time supervision, audio/video quality analysis (EBU R128 loudness, freeze, black), SCTE-35 metadata on an operator timeline, circular replay and GPIO/Ember+ automations. One console. No blind spots. Faster decisions.
One probe to supervise your entire broadcast chain
Reduce diagnosis time by 70%. SondaStream (Sonda Stream) brings media streams, SCTE-35 markers, audio/video quality (EBU R128, ETR 101 290), replay and GPIO/Ember+ automations into one console. Your NOC reacts faster while engineering keeps full technical depth.
Dense supervision, no useless noise.
Degraded streams stay on top. Status, alerts, and trends remain immediately readable whether you monitor a few critical services or a full fleet.
- Configurable Streams console: matrix, table, NOC view, and status/type filters.
- Compact global view: recent incidents, unstable streams, and health score.
- Contextual navigation: from weak signal to impacted stream, then replay or action.
- Mobile companion: search, alerts, and quick field consultation.
Less time wasted searching. The console prioritizes operations first, then adds the depth needed for investigation.
From status to diagnosis in one workspace.
Each stream gets a consistent workspace: live player, circular replay, quality analysis, events, transport inspection, and automations. Signals are not just collected, they are correlated to help operators interpret situations quickly.
Playlist integrity, stalls, duplicates, discontinuities, and rendition validation.
ETR 101 290 P1, PCR jitter, CC errors, PAT/PMT, SCTE-35 extraction.
Silence, LUFS loudness (EBU R128), true peak dBTP, dBFS measurements.
Black frame and freeze detection with thresholds and time tolerance.
Automatic correlation: transport errors, quality degradations, and SCTE-35/ID3 events are synchronized on one timeline for faster diagnosis.
Recover incident context in seconds.
Event log global
Aggregated history, server-side filters, interpreted inspection, and raw payload: operators keep a clear view even when event volume increases.
Multi-stream circular recorder
Indexed replay, time navigation, asynchronous extraction, and SHA-256 fingerprinting: investigation relies on usable archive evidence, not just an alert.
The timeline that links program, ads, quality, and field actions.
SondaStream interprets SCTE-35 messages to rebuild an operator-usable business view: program phases, ad placements, and transitions visible directly in the operator timeline.
- Contextual SCTE-35 interpretation: raw events become a readable program/ad timeline.
- I-frame image validation: insertion alignment is checked to avoid visible cuts.
- Multi-rail correlation: SCTE-35, audio/video quality, transport, and diagnostics on one time window.
- Triggered automations: GPIO and Ember+ actions driven by SCTE messages with stream-scoped rules.
- Execution traceability: every decision and action remains visible in event history.
Editorial segments identified and tracked across the active range.
Ad windows detected with immediate operator readability.
Temporal consistency check before automated action.
Reliable, logged, and auditable field triggering.
Turn broadcast markers into field actions.
Rules are configured per stream. One automation can combine multiple operations and also handle event end logic. Every action is validated, logged, and traceable.
GPIO
Brainboxes ED-008, ED-527, ED-538 control with real-time state.
Ember+
Lawo, Grass Valley, Riedel, VSM: parameters and matrix routing.
SCTE-35
time_signal, splice_insert, segmentation descriptors enrichis.
Typical use cases: local control-room GPI trigger on break_start, automatic source switch on network_start, dynamic audio matrix routing on program_end, recorder marker on advertisement.
Credible media supervision starts with an observable probe.
CPU, memory, I/O pressure, disk, network, PostgreSQL, recorders, and extraction jobs are visible directly in the interface.
Continuous monitoring and automatic reconnection.
CIFS / NFS, mount health checks, and fail-safe blocking.
Synchronization managed directly from the console.
Scoped keys, security controls, and integrations.
Deployable in real operations. The probe tracks its resources, services, and telemetry freshness.
Connect supervision to existing broadcast workflows.
"An operator console, a recorder, an I/O workbench, and an integration API: SondaStream, also searched as Sonda Stream, brings together what used to be fragmented."
Audio and video HLS (LL-HLS)
UDP / MPEG-TS multicast
Icecast / HTTP audio (MP3, AAC, Opus)
AES67 / RTP PCM (L16, L24)
RTMP, RTMPS, and SRT via live bridge
Automatic codec and rendition detection
Full SCTE-35 (time_signal, splice_insert, segmentation)
Binary ID3: title, artist, artwork
HLS timed metadata
API metadata with bearer token
ETR 101 290 P1: sync, PAT, PMT, CC
LUFS loudness (EBU R128, ATSC A/85)
True peak dBTP, dBFS
Silence, freeze, black frame
Full REST API (OpenAPI / Swagger)
Real-time WebSocket (alerts, SCTE-35, quality)
MCP server: 80+ tools for AI and CLI
Brainboxes GPIO (ED-008, ED-527, ED-538)
Ember+ (Lawo, Grass Valley, Riedel, VSM)
AWS IAM Identity Center SSO (SAML 2.0)
Health checks for external monitoring
JSON configuration export/import
Visit the demo probe
An online instance is available to explore the main workspaces: Streams console, SCTE-35 timeline, I/O workbench, recorder, and quality analysis.
URL https://probe.jjtronics.com/ ↗
Login probe Password Sonda.Stream
Everything the probe brings to operations.
Real-time supervision
- Continuous monitoring or configurable polling.
- Real-time WebSocket with event latency below 2 seconds.
- Automatic reconnection with exponential backoff.
- Graceful shutdown and explicit probe-unavailable periods.
- Operator health score, confidence, and correlation notes.
- Dense Streams console: matrix, table, NOC view, filters, presets, and configurable columns.
- Compact Overview for NOC triage.
- Aggregated Alerts / Events console with server-side filtering and pagination.
- Activity console for recent operations.
- Contextual navigation to stream, replay, rule, or related job.
- Responsive UI with companion mobile navigation.
- Per-stream configurable alerting with custom thresholds.
- Event history with raw payload and operator interpretation.
Protocols and playback
- Audio and video HLS (Apple HLS, LL-HLS).
- UDP / MPEG-TS multicast and unicast.
- Icecast / HTTP audio (MP3, AAC, Opus, Vorbis).
- AES67 / RTP PCM audio (L16, L24).
- Live HLS bridge for UDP, RTP, RTMP, RTMPS, and SRT.
- Operator player: audio waveform, real-time VU meter, ID3 now-playing, HLS quality selection, multilingual audio tracks, WebVTT subtitles, and media panel.
- DVB teletext subtitle bridge to WebVTT with multi-page support.
- Automatic audio/video codec detection.
- Support for multiple audio tracks and alternative HLS renditions.
Quality and metadata
- HLS integrity checks: stalls, duplicates, regressions, sequence jumps, and discontinuities.
- Transport continuity at HLS segment level with CRC validation.
- TS analysis via TSDuck and ETR 101 290 priority-1 checks: TS sync loss, sync byte error, PAT error, continuity count error, PMT error, PID error.
- SCTE-35 detection: time_signal, splice_insert, enriched segmentation descriptors (program_start, program_end, break_start, break_end, provider_advertisement_start, distributor_advertisement_start, placement_opportunity_start, network_start).
- Real-time binary ID3 parser: title, artist, genre, album, and artwork with APIC support.
- Per-stream API metadata polling with bearer token, custom headers, and automatic retries.
- Dedicated timeline API with interpreted operator timestamps and event correlation.
- Audio QC: silence detection, LUFS loudness (EBU R128, ATSC A/85), true peak dBTP, dBFS measurements.
- Video QC: freeze frame and black frame with configurable thresholds and temporal tolerance.
- One-shot deep analysis with structured JSON results, bitrate statistics, and integrity report.
- Automatic correlation of transport errors, audio/video quality, and metadata on the timeline.
SCTE-35 timeline and interpretation
- Operator timeline with Program and Ad rails.
- Contextual interpretation of SCTE-35 messages into readable business phases.
- Automatic I-frame alignment validation for ad insertions.
- Temporal inconsistency detection (out-of-order, overlaps).
- Synchronized view of SCTE-35 events, audio/video quality, and transport diagnostics.
- Support for immediate and pre-announced events with countdown.
- Full traceability: raw SCTE-35 message, operator interpretation, and triggered actions.
- Timeline export to JSON for analysis or archiving.
Recorder, automations, security, and integration.
Recorder and storage
- Multi-stream circular recorder with indexed chunks and PostgreSQL metadata.
- Time-based replay with Errors, SCTE-35, ID3, Timed Metadata, Diagnostics, and Chunks rails.
- Asynchronous extraction with SHA-256 fingerprint, real-time progress, and notifications.
- Export formats: native HLS playlists, native MPEG-TS, isolated audio segments.
- Native HLS archiving with playlist metadata preservation (codecs, bandwidths, resolutions).
- Native Upipe recorder for UDP / MPEG-TS with low latency.
- Native GStreamer audio segmentation for Icecast and AES67 with precise chunking.
- CIFS / NFS NAS management: share discovery, automatic mount, mount health checks, and disconnect alerts.
- Fail-safe recorder blocking when expected storage is not actually mounted (prevents local fallback writes).
- Extraction jobs with queue, history, automatic retries, and scheduled deletion.
- Storage tracking in console: used/free space, quota, and retention projection.
- Intelligent per-stream circular retention with exported excerpt preservation.
Automations and I/O
- Per-stream SCTE-35 rules with conditional logic.
- Full support for time_signal and enriched segmentation descriptors (program, break, advertisement, placement, network, chapter, unscheduled, distributor, provider).
- Multiple triggers: program_start, program_end, break_start, break_end, advertisement, placement_opportunity, network_start, network_end.
- One rule can combine several actions: Brainboxes GPIO, Ember+ parameter write, Ember+ matrix routing.
- Optional event-end actions (GPIO auto-reset, Ember+ output parameter).
- Filtering by upid_type, segmentation_type_id, and delivery_not_restricted.
- Central I/O workbench for Brainboxes (ED-008, ED-527, ED-538) and Ember+ (Lawo, Grass Valley, Riedel, VSM).
- Live GPIO state (input/output), Ember+ browser, and chronological per-device logs.
- Manual GPIO and Ember+ action testing from the console with preview before execution.
- I-frame validation before triggering actions to avoid visible cuts.
- Execution logs with timestamp, triggered rule, source SCTE-35 message, and action result.
Authentication and security
- Local authentication with bcrypt hashing and salting.
- SSO support via AWS IAM Identity Center (SAML 2.0).
- User management: create, edit, disable, granular roles and permissions.
- Roles: Admin, Operator, Viewer with resource-level access control.
- Scoped API keys: full web API, MCP server, recorder access, read-only.
- Immediate API key revocation and deletion.
- User sessions with configurable expiration and automatic renewal.
- HTTPS / TLS support with Let's Encrypt or custom certificates.
- CSRF protection, input validation, and SQL sanitization.
- Audit logs for sensitive actions (user creation, rule changes, recorder extraction).
Operations and integration
- Resources page: CPU, RAM, swap, I/O pressure, disk (usage, inodes), network (rx/tx, errors), and telemetry freshness.
- Service tracking: web API, PostgreSQL, active recorders, extraction jobs, and workers.
- Pause analysis without stopping recorder (maintenance mode).
- Per-stream configurable event retention with automatic purge.
- Settings console: operations, retention, alerting, NAS storage, security, and NTP synchronization.
- API keys with custom names, multiple scopes, and last-used date.
- MCP server (Model Context Protocol) with 80+ tools for AI assistants (Claude Desktop, Cursor), CLI, and programmatic integrations.
- Full REST API with interactive OpenAPI / Swagger docs.
- Real-time WebSocket events: alerts, status changes, SCTE-35, ID3, audio/video quality.
- JSON backup/import for configuration: streams, rules, GPIO, Ember+, users (without passwords), global settings.
- NTP settings with multiple servers, sync validation, and drift alerts.
- Probe restart and shutdown with countdown and connected-user notification.
- Batch editor for fleet-scale operations: enable/disable, retention changes, rule updates.
- Icecast delay comparison through audio correlation with confidence score and millisecond delta.
- Multicast Discovery: automatic IP/port range scan with active stream detection.
- Playout Stream Loop: looped playback of media files from NAS, automatic probing, multicast output, and configurable autostart.
- Health check endpoint for external monitoring integrations (Prometheus, Nagios, Zabbix).
- Structured JSON logs with automatic rotation and configurable levels (debug, info, warn, error).
- Docker deployment support with persistent PostgreSQL and storage volumes.
The SondaStream project is built by a broadcast geek.
I work at M6/RTL and face real operational constraints every day. SondaStream (Sonda Stream) was born from that field reality: less noise, more useful signal, and tools that truly help NOC and engineering teams decide fast.
Is SondaStream related to SodaStream?
No. SondaStream is a broadcast probe. The word sonda comes from Latin and refers to probing, measurement and technical observation.
When did the project start?
The repository started on June 6, 2025. The idea had been in my head for years before it became SondaStream.
Which protocols are monitored natively?
HLS, UDP/MPEG-TS, Icecast and AES67, with correlation across transport, quality, SCTE-35 and ID3 events.
Which media engines does it use?
SondaStream combines FFmpeg/ffprobe, TSDuck, Upipe, and GStreamer depending on the protocol: TS and SCTE-35 analysis, ID3 metadata, UDP/MPEG-TS recording, Icecast, and AES67.
Who is SondaStream for?
NOC, operations and broadcast engineering teams that need real-time visibility and faster diagnosis.
Is it designed for real operations?
Yes. The workflow is built for the field: triage, replay-based evidence and automated I/O actions.