Mcaster1DSPEncoder — Refactor & Development Roadmap

Project Overview

Mcaster1DSPEncoder is an open-source, multi-format audio streaming DSP encoder suite — a maintained, modernized fork of the EdCast/Oddsock DSP encoder lineage. It ships as a plugin for Winamp, foobar2000, and RadioDJ, and as a self-contained encoder application targeting Shoutcast and Icecast servers.

Phase 1 — VS2022 Build Fix COMPLETE

All issues blocking compilation under Visual Studio 2022 (v143 toolset, Windows 10 SDK) were resolved. All 4 build targets compile cleanly and run without crashes.

Fixes Applied

• foobar2000_SDK / helpers: Updated _WIN32_WINNT=0x501 → 0x0601 to resolve _REASON_CONTEXT C2011 struct redefinition against Win10 SDK
• pfc/alloc.h: Fixed alloc_fixed::alloc::move_from() calling pfc::copy_array_t() on a type lacking has_owned_items — replaced with element-by-element copy loop
• foobar2000/SDK/guids.cpp: Overrode FOOGUIDDECL (__declspec(selectany)) with empty define so class_guid symbols become regular externals visible to lib.exe, resolving mass LNK2001
• All vcxproj files: Unified CRT/MFC runtime to consistent /MD (MultiThreadedDLL), eliminating STATUS_STACK_BUFFER_OVERRUN (0xC0000409) crash from mixed /MT+/MD build

Phase 2 — Rebrand & Refactor COMPLETE

Goal: Complete renaming of every AltaCast identity in the codebase. No new features. "Standalone" is dropped from all names — the application is self-contained by definition.

2.1 Naming Conventions

2.2 Deep Naming Audit — All 150+ Instances

Complete inventory of all instances requiring change, organized by category.

All references in *.iss installer script:

• App name, app publisher URL, output base filename
• Source: file paths for dsp_altacast.dll, foo_altacast.dll, altacastStandalone.exe
• [Registry] keys using AltaCast as a key name

2.3 File & Directory Renames

Ordered sequence to avoid broken references mid-rename:

Step 1 — Rename source files inside libaltacast/:
  libaltacast.cpp         → libmcaster1dspencoder.cpp
  libaltacast.h           → libmcaster1dspencoder.h
  libaltacast_socket.cpp  → libmcaster1dspencoder_socket.cpp
  libaltacast_socket.h    → libmcaster1dspencoder_socket.h
  libaltacast_resample.h  → libmcaster1dspencoder_resample.h
  libaltacast.vcxproj     → libmcaster1dspencoder.vcxproj

Step 2 — Rename libaltacast/ folder:
  src/libaltacast/        → src/libmcaster1dspencoder/

Step 3 — Rename plugin source files:
  altacast_winamp.cpp     → mcaster1_winamp.cpp
  altacast_winamp.h       → mcaster1_winamp.h
  altacast_foobar.cpp     → mcaster1_foobar.cpp
  altacast_radiodj.cpp    → mcaster1_radiodj.cpp
  altacast_radiodj.h      → mcaster1_radiodj.h

Step 4 — Rename shared files:
  altacast.h              → mcaster1dspencoder.h
  altacast.rc             → mcaster1dspencoder.rc
  altacastStandalone.h    → Mcaster1DSPEncoder.h
  altacastStandalone.cpp  → Mcaster1DSPEncoder.cpp

Step 5 — Rename project files:
  altacastStandalone.vcxproj  → Mcaster1DSPEncoder.vcxproj
  altacast_winamp.vcxproj     → mcaster1_winamp.vcxproj
  altacast_foobar.vcxproj     → mcaster1_foobar.vcxproj
  altacast_radiodj.vcxproj    → mcaster1_radiodj.vcxproj

Step 6 — Rename solution files:
  altacastStandalone.sln  → Mcaster1DSPEncoder.sln
  altacast_winamp.sln     → mcaster1_winamp.sln
  altacast_foobar.sln     → mcaster1_foobar.sln
  altacast_radiodj.sln    → mcaster1_radiodj.sln

2.4 Credits & Attribution

In all About dialogs, splash screens, installer text, and source file headers:

Mcaster1 DSP Encoder
Maintainer: Dave St. John <davestj@gmail.com>
https://github.com/davestj/Mcaster1DSPEncoder

Based on EdCast/Oddsock DSP encoder by Ed Zaleski
Original project: https://www.oddsock.org/

All source file copyright comment blocks — ADD, do not replace original:

// Mcaster1DSPEncoder — fork of EdCast/Oddsock DSP encoder
// Maintainer: Dave St. John <davestj@gmail.com>
// Original Author: Ed Zaleski

2.5 MP3 — Replace BladeMP3 with Native LAME

AltaCast uses BladeMP3EncDLL.h — an obsolete Windows-only interface. All US MP3 patents expired April 23, 2017. LAME (LGPL) is now freely distributable. LAME 3.100 source is in external/lame/; also available via vcpkg mp3lame:x64-windows.

1. Upgrade external/lame/vc_solution/vc9_libmp3lame_dll.vcproj to v143, build libmp3lame.dll
2. Replace #ifdef WIN32 BladeMP3 guards with #ifdef HAVE_LAME using standard lame.h API
3. Remove BladeMP3EncDLL.h from external/include/
4. Distribute libmp3lame.dll with all targets

2.6 Dual-Platform Build: x86 (Win32) and x64

Goal: Ship both 32-bit and 64-bit builds to maximize Windows compatibility. x64 for modern systems (Windows 10/11); x86 for legacy environments and 32-bit host applications (older Winamp builds, 32-bit RadioDJ, etc.). DSP plugins must match the bitness of the host application, so shipping both ensures broadest compatibility.

Step 2 — Add x64 Configurations to All vcxproj Files

Each project currently has Win32|Debug and Win32|Release. Add x64|Debug and x64|Release to each. Key changes per x64 config group:

• <Platform>Win32</Platform> → <Platform>x64</Platform>
• Linker: MachineX86 → MachineX64
• vcpkg paths: x64-windows\ for x64 configs, x86-windows\ for Win32 configs
• Output dirs already vary by config — ensure x64\Release\ and Win32\Release\ are distinct

Step 3 — DRY vcpkg Paths via MSBuild Property

<PropertyGroup>
  <VcpkgTriplet Condition="'$(Platform)'=='Win32'">x86-windows</VcpkgTriplet>
  <VcpkgTriplet Condition="'$(Platform)'=='x64'">x64-windows</VcpkgTriplet>
  <VcpkgInclude>$(VCPKG_ROOT)\installed\$(VcpkgTriplet)\include</VcpkgInclude>
  <VcpkgLib>$(VCPKG_ROOT)\installed\$(VcpkgTriplet)\lib</VcpkgLib>
</PropertyGroup>

Step 4 — Release Artifact Layout

dist/
  Win32/
    Mcaster1DSPEncoder.exe
    dsp_mcaster1.dll  dsp_mcaster1_radiodj.dll  foo_mcaster1.dll
    libmp3lame.dll  opus.dll  opusenc.dll  fdk-aac.dll  ...
  x64/
    Mcaster1DSPEncoder.exe
    dsp_mcaster1.dll  dsp_mcaster1_radiodj.dll  foo_mcaster1.dll
    libmp3lame.dll  opus.dll  opusenc.dll  fdk-aac.dll  ...

Compatibility Matrix

2.7 Build System Cleanup

• Update build_all.ps1: build both Win32 and x64 for all targets, copy runtime DLLs from correct vcpkg triplet per platform, populate dist/Win32/ and dist/x64/
• Remove build.ps1 and build_winamp.ps1 (consolidated into build_all.ps1)
• Add .gitignore: Release/, Debug/, x64/, Win32/, dist/, *.user, *.suo, *.aps, *.pch, *.obj, *.ilk, ipch/, .vs/, *.log, *.tlog
• Initialize fresh git repository in project root

2.8 Phase 2 Deliverables

• All source files, folders, and projects renamed per Section 2.3
• All 150+ string/symbol instances replaced per Section 2.2 deep audit
• All 4 targets compile clean (x86 and x64) with zero AltaCast references remaining
• x86-windows vcpkg triplets installed for all required packages
• x64 configurations added to all vcxproj files alongside existing Win32 configs
• VcpkgTriplet MSBuild property wires correct include/lib paths per platform
• About dialogs show correct attribution (Ed Zaleski original, davestj maintainer)
• Mcaster1DSPEncoder.exe launches and encodes — both x86 and x64 builds
• dsp_mcaster1.dll loads in Winamp — both x86 and x64 builds
• foo_mcaster1.dll loads in foobar2000 — both x86 and x64 builds
• dsp_mcaster1_radiodj.dll loads in RadioDJ — both x86 and x64 builds
• Native LAME MP3 via libmp3lame.dll working, BladeMP3 removed
• dist/Win32/ and dist/x64/ populated with correct runtime DLLs
• build_all.ps1 builds both platforms in one pass
• README.md updated
• .gitignore in place
• Initial git commit tagged v0.2.0

2.9 ResizableLib & Dialog Modernization COMPLETE

All MFC dialogs were modernized with resizable layout and professional flat styling using ResizableLib (Paolo Messina, Artistic License) from external/ResizableLib/. This was completed as part of Phase 2 to establish a consistent UI foundation before audio codec work.

Changes Applied

Dialog Style Updates (mcaster1dspencoder.rc)

• DS_MODALFRAME → WS_THICKFRAME on all resizable dialogs (enables drag-resize)
• Font updated to 9, "Segoe UI", 400, 0, 0x1 throughout (was 8, "MS Sans Serif")
• DS_3DLOOK removed for flat look; WS_EX_COMPOSITED NOT used (RC compiler does not support it)
• About dialog retains DS_MODALFRAME by design (fixed-size, no resize needed)

Build Integration — All 4 Targets

ResizableLib sources compiled into all 4 build targets (standalone EXE + 3 DLL plugins), since all plugin projects compile the full MFC dialog source set. Each target's vcxproj has 8 ResizableLib ClCompile entries with:

<PrecompiledHeader>NotUsing</PrecompiledHeader>
<AdditionalOptions>/wd4005 %(AdditionalOptions)</AdditionalOptions>
<AdditionalIncludeDirectories>../external/ResizableLib;...</AdditionalIncludeDirectories>

NotUsing allows ResizableLib to find its own StdAfx.h (which defines WINVER=0x0501) via directory-first include resolution. /wd4005 suppresses the WINVER redefinition warning against the project-level 0x0601.

Phase 3 — Audio Backend Modernization COMPLETE

Goal: Remove WMA (end-of-life). Add dual MP3 paths (LAME built-in + ACM), Opus, and full HE-AAC (AAC-LC / HE-AACv1 / HE-AACv2) via fdk-aac. Replace bass.dll capture with statically-linked PortAudio (WASAPI / ASIO / MME / DS / WDM-KS). Achieve format parity with RadioBOSS: MP3 (×2), AAC-LC, AAC+, AAC++, Ogg Vorbis, Opus, FLAC.

vcpkg Packages for Phase 3

3.1 Remove basswma.dll / WMA REMOVE

Microsoft deprecated Windows Media Services. basswma.dll is a proprietary Un4seen plugin and is EOL.

• Remove basswma.dll/.lib from external/lib/
• Remove basswma.h from external/include/
• Remove all WMA encoder code paths from libmcaster1dspencoder.cpp
• Remove WMA UI controls from all plugin config dialogs
• Remove basswma.lib from all vcxproj linker inputs

3.2 Add Opus Encoder

vcpkg provides opus + libopusenc ready to link. Integration via libopusenc high-level API:

#include <opusenc.h>

OggOpusComments *comments = ope_comments_create();
ope_comments_add(comments, "ENCODER", "mcaster1dspencoder");
OggOpusEnc *enc = ope_encoder_create_callbacks(
    &callbacks, userdata, comments, 48000, channels, 0, &error);
ope_encoder_ctl(enc, OPUS_SET_BITRATE(bitrate * 1000));
ope_encoder_ctl(enc, OPUS_SET_COMPLEXITY(10));
ope_encoder_write(enc, pcm, samples_per_channel);
ope_encoder_drain(enc);
ope_encoder_destroy(enc);
ope_comments_destroy(comments);

MIME: audio/ogg; codecs=opus. Mount point extension: .opus.

3.3 Add AAC and AAC+ (HE-AAC) via fdk-aac

Replace the legacy libfaac dependency with fdk-aac 2.0.3, which supports AAC-LC, HE-AAC (AAC+), and HE-AAC v2 (AAC++/eAAC+). This achieves full RadioBOSS format parity.

Format Matrix

Integration via fdk-aac C API

#include <fdk-aac/aacenc_lib.h>

HANDLE_AACENCODER hEncoder;
aacEncOpen(&hEncoder, 0, channels);

// Profile selection:
// AOT_AAC_LC (2)  = standard AAC-LC
// AOT_SBR    (5)  = HE-AAC v1 (AAC+)  — adds Spectral Band Replication
// AOT_PS     (29) = HE-AAC v2 (AAC++) — adds Parametric Stereo on top of SBR
aacEncoder_SetParam(hEncoder, AACENC_AOT,         AOT_SBR);   // AAC+ example
aacEncoder_SetParam(hEncoder, AACENC_SAMPLERATE,  44100);
aacEncoder_SetParam(hEncoder, AACENC_CHANNELMODE, MODE_2);
aacEncoder_SetParam(hEncoder, AACENC_BITRATE,     64000);
aacEncoder_SetParam(hEncoder, AACENC_TRANSMUX,    TT_MP4_ADTS); // ADTS for live streaming

aacEncEncode(hEncoder, &inBufDesc, &outBufDesc, &inArgs, &outArgs);
aacEncClose(&hEncoder);

Recommended Bitrates (RadioBOSS Parity)

3.4 Replace bass.dll with PortAudio

• Remove bass.lib from Winamp / foobar2000 / RadioDJ linker deps (plugins intercept audio via host; no device capture needed)
• Integrate PortAudio v19 for Mcaster1DSPEncoder.exe device capture path
• PortAudio provides device enumeration and PCM stream capture with WASAPI / DirectSound backend
• License: MIT — freely distributable, static or dynamic linkage

3.5 RadioBOSS Format Parity Summary

3.6 Phase 3 Deliverables

• basswma.dll removed, WMA code paths deleted
• fdk-aac:x64-windows[he-aac] installed via vcpkg and linked
• AAC-LC encoder integrated in all 4 targets
• AAC+ (HE-AAC v1) encoder integrated in all 4 targets
• AAC++ (HE-AAC v2) encoder integrated in all 4 targets
• Legacy libfaac removed (replaced by fdk-aac)
• Opus encoder integrated in all 4 targets
• Opus Ogg stream verified in Icecast and VLC
• AAC ADTS stream verified in Icecast and compatible players
• Format config UI updated in all plugin dialogs (AAC-LC / AAC+ / AAC++ dropdown)
• bass.lib removed from plugin linker deps
• PortAudio integrated for device capture in Mcaster1DSPEncoder.exe
• Redistributables updated: libmp3lame.dll, opus.dll, opusenc.dll, fdk-aac.dll

Phase 4 — YAML Multi-Station Configuration COMPLETE

Goal: Replace legacy flat key=value config with structured YAML supporting multiple simultaneous stations, per-station encoder chains, ICY/YP metadata, and save-to-WAV.

4.1 Current Config Problems

• No station grouping concept — encoder slots that logically belong together (same server, same ICY metadata, different formats) have no way to express that relationship in config
• Duplicated metadata — ICY name, genre, URL, password must be copy-pasted into every numbered file for encoders targeting the same server
• Opaque numbering — _1.cfg vs _2.cfg carries no semantic meaning; figuring out which slot goes where requires reading every file
• Hard cap of 10 slots (MAX_ENCODERS 10) — increasing requires a recompile
• Flat key namespace — no grouping or structure within each file; no inline comments describing what the encoder is for
• No archive configuration — no built-in save-to-WAV settings per encoder slot

4.2 New YAML Config Format

File: mcaster1dspencoder.yaml

global:
  log_level: info
  log_file: mcaster1dspencoder.log
  reconnect_delay: 5
  reconnect_on_drop: true

stations:
  - id: 1
    name: "My Rock Station"
    enabled: true
    icy:
      name: "My Rock Station"
      genre: "Rock"
      url: "https://mystation.example.com"
      description: "The best rock on the internet"
      pub: true
    server:
      type: icecast2
      host: stream.example.com
      port: 8000
      password: "hackme"
      mount: /rock
    encoders:
      - format: mp3
        bitrate: 128
        samplerate: 44100
        channels: 2
        quality: 5
        vbr: false
      - format: ogg
        bitrate: 96
        samplerate: 44100
        channels: 2
        quality: 0.6
      - format: opus
        bitrate: 128
        samplerate: 48000
        channels: 2
        complexity: 10
        signal: audio
      - format: aac
        profile: he-aacv1     # lc | he-aacv1 | he-aacv2
        bitrate: 64
        samplerate: 44100
        channels: 2
    archive:
      enabled: false
      path: "./recordings/"
      filename_pattern: "{station_id}_{date}_{time}.wav"
      max_size_mb: 2048

  - id: 2
    name: "My Jazz Stream"
    enabled: false
    icy:
      name: "Smooth Jazz 24/7"
      genre: "Jazz"
      url: "https://jazz.example.com"
      pub: false
    server:
      type: shoutcast2
      host: shout2.example.com
      port: 8002
      password: "jazzpass"
      mount: /jazz
    encoders:
      - format: mp3
        bitrate: 192
        samplerate: 44100
        channels: 2
        quality: 2
        vbr: true
        vbr_min: 96
        vbr_max: 320
    archive:
      enabled: true
      path: "D:/archive/jazz/"
      filename_pattern: "jazz_{date}.wav"
      max_size_mb: 4096

4.3 YAML Library — libyaml (already in vcpkg)

New module: libmcaster1dspencoder/config_yaml.cpp + config_yaml.h

int mcaster1_load_yaml_config(const char *path, mcaster1_config_t *cfg);
int mcaster1_get_station_count(const mcaster1_config_t *cfg);
int mcaster1_get_station(const mcaster1_config_t *cfg, int idx, mcaster1_station_t *st);
int mcaster1_save_yaml_config(const char *path, const mcaster1_config_t *cfg);

Legacy key=value reader retained as fallback. Format auto-detected by file extension.

4.4 Multi-Station Architecture

MAX_ENCODERS increased from 10 → 32. Each YAML station encoder entry maps to one mcaster1Globals slot. Audio input is fanned out to all active encoder threads from a single capture/input callback using the existing pthreadVSE threading model.

4.5 Phase 4 Deliverables

• libyaml linked from vcpkg (x86-windows + x64-windows, static)
• config_yaml.cpp / config_yaml.h module in libmcaster1dspencoder
• mcaster1dspencoder.yaml sample config with 2 stations
• YAML loader replaces legacy cfg loader (legacy kept as fallback, auto-detected by extension)
• MAX_ENCODERS increased 10 → 32
• Multi-station verified: 2 stations, 2+ encoders each, simultaneous encode
• Per-station ICY metadata verified on Icecast and Shoutcast
• Save-to-WAV archive functional per station with filename pattern templating
• Station list UI in Mcaster1DSPEncoder.exe
• Station selector in all plugin config dialogs

Phase 5 — Podcast Feature Set & ICY 2.2 Metadata IN PROGRESS

Goal: Transform the encoder into a first-class podcast platform. When a DJ saves a live stream to a local file it becomes an archived podcast episode. Phase 5 adds a dedicated Podcast Settings UI tab, automatic RSS 2.0 + iTunes feed generation alongside each saved recording, and full ICY 2.2 extended header injection on every Icecast2 / Mcaster1DNAS connect. Config save hardening ensures settings are never lost on exit.

5.1 Podcast Settings & Archive Recording

The existing “Advanced Settings” config tab has been redesigned as a proper Podcast Settings tab (IDD_PROPPAGE_LARGE2). Log settings are now in their own group box at the bottom. New controls:

• Generate RSS feed alongside saved file — checkbox (IDC_GENERATE_RSS)
• Populate RSS from YP Settings — auto-fills podcast fields from YP tab (IDC_RSS_USE_YP)
• Browse button for archive directory (IDC_BROWSE_ARCHIVE)
• Podcast Title, Author/Host, Category, Language, Copyright, Website URL, Cover Art URL, Description (multiline)

Field resolution order (first non-empty wins): Podcast Settings → YP Settings → safe defaults.

5.2 RSS 2.0 + iTunes Feed Generation

New files src/podcast_rss_gen.h and src/podcast_rss_gen.cpp implement a pure-C RSS writer with zero extra dependencies (pure fprintf). Triggered automatically from closeArchiveFile() when Generate RSS is enabled.

5.3 ICY 2.2 Extended Header Integration

A new 4th config tab “ICY 2.2 Extended” (IDD_PROPPAGE_LARGE3) exposes all 12 ICY 2.2 header categories from the ICY 2.2 spec. On every Icecast2 SOURCE connect, appendICY22Headers() appends all non-empty fields after the standard ICE headers.

The session-id is auto-generated as a UUID via CoCreateGuid() on the first connect if the field is blank. Shoutcast servers silently ignore the unknown headers — safe to send always.

5.4 Config Save Hardening

• OnClose() — writeMainConfig() called before exit(1) to flush config on every close path
• CleanUp() — Pa_Terminate() added to pay the PortAudio shutdown debt
• ProcessConfigDone() — writeMainConfig() added so NumEncoders + window state persists whenever the encoder OK button is clicked, not only on app exit

5.5 Phase 5 Deliverables

• Podcast Settings tab rework (IDD_PROPPAGE_LARGE2) — two group boxes, Browse button, RSS checkboxes, 8 podcast fields
• ICY 2.2 Extended tab (IDD_PROPPAGE_LARGE3) — ~55 fields across 10 header categories
• podcast_rss_gen.cpp/.h — RSS 2.0 + iTunes NS feed generator (pure C)
• ICY22Settings.cpp/.h — full MFC dialog for 4th config tab, UUID auto-gen
• libmcaster1dspencoder.h — 11 podcast + ~55 ICY 2.2 fields added to mcaster1Globals
• libmcaster1dspencoder.cpp — path capture in openArchiveFile(), RSS call in closeArchiveFile(), appendICY22Headers(), Icecast2 integration
• config_yaml.cpp — 75+ new YAML keys for podcast + ICY 2.2 round-trip persist
• Config.cpp/.h — 4th tab wired; GlobalsToDialog/DialogToGlobals extended
• MainWindow.cpp — 3 config save hardening fixes
• All 3 main vcxproj files updated with new source files
• Pending: build verification + E2E test (RSS file, ICY 2.2 headers in DNAS log)

Phase 6 — Mcaster1DNAS Integration PLANNED

Goal: First-class integration between Mcaster1DSPEncoder and the Mcaster1DNAS streaming server. They are already ecosystem companions — this phase makes them work together seamlessly as a matched pair, better together than either is with generic third-party tools.

Integration Features

Phase 6 Deliverables

• LAN auto-discovery: mDNS/UDP broadcast — populate server picker from discovered DNAS instances
• Shared YAML config schema v2 — station definitions consumed by both encoder and DNAS
• Live stream health API polling — listener count + bitrate in encoder UI per slot
• Song history push on track change — direct ring buffer API call, zero polling delay
• Shared admin API key — single credential in YAML, used by both apps
• Relay/fallback trigger — encoder signals DNAS relay switch on source drop/recover
• Verified end-to-end: Encoder connected to DNAS, Track History pages updating in real time

Phase 7 — Mcaster1Castit PLANNED

Goal: Resurrect and modernize Castit — the original live broadcast scheduling and automation tool built for casterclub.com in 2001 with Ed Zaleski's guidance. Castit was written against VC6 and will be upgraded to Visual Studio 2022 as a first-class member of the Mcaster1 ecosystem.

Upgrade Plan

Mcaster1Castit Feature Goals

• Playlist scheduling: time-based and event-triggered playback with calendar UI
• Live source injection: interrupt playlist for a live broadcast mid-schedule
• Integration with Mcaster1DNAS for stream status and listener statistics
• Integration with Mcaster1DSPEncoder for encoding, stream health, and format selection
• ICY 2.1 metadata injection from the playlist schedule on each track change
• Archive recording: save each scheduled slot to WAV/MP3/FLAC for replay and archival
• REST API: remote schedule management via JSON API

Phase 7 Deliverables

• Castit VC6 project upgraded to VS2022 — compiles with v143 toolset
• libxml2 integrated via vcpkg replacing bundled VC6 build
• libcurl integrated via vcpkg with TLS 1.3 support
• libmariadb (or libmysql) integrated via vcpkg
• ResizableLib applied to all Castit dialogs (matching Mcaster1DSPEncoder style)
• YAML configuration replacing legacy flat config
• Mcaster1DSPEncoder encoding engine integrated for live source encoding
• ICY 2.1 metadata injection on track change events
• Mcaster1Castit.exe launches, schedules, and streams to a local Mcaster1DNAS instance

Phase 8 — Analytics, Metrics & Platform Engagement PLANNED

Goal: Leverage the ICY 2.1 protocol and the Mcaster1DNAS song history ring buffer to build a statistical metrics and analytics layer that gives station operators actionable listener engagement data — without requiring third-party analytics services or sharing data with external platforms.

Analytics Architecture

[Listeners] ──ICY 2.1 engagement signals──► [Mcaster1DNAS]
                                                    │
                                          [Analytics Ring Buffer]
                                                    │
                           ┌────────────────────────┼──────────────────────────┐
                           ▼                        ▼                          ▼
                  [Track Play Stats]        [Listener Peaks]          [Format Stats]
                  Most played, rotation     Hour-by-hour curve        Codec popularity
                           │
                  [Mcaster1Castit Scheduler]
                           │
                  [Optimize playlist rotation
                   based on engagement data]

Planned Metrics

Platform Engagement via ICY 2.1

The ICY 2.1 metadata block includes a history_url field pointing to the DNAS track history API. Listener-side players that support ICY 2.1 can display:

• A clickable "now playing" card with album art and music service links
• A scrollable track history for the current listening session
• An "add to playlist" action via Spotify / Apple Music deep link
• A listener engagement button (like/react) that feeds back into DNAS analytics

This creates a direct engagement path from the audio stream to music service platforms, measurable through history_url click analytics on the DNAS side — without requiring a third-party ad network or analytics vendor.

Mcaster1Castit Playlist Intelligence

Phase 8 closes the loop: analytics data collected by Mcaster1DNAS feeds back into Mcaster1Castit's scheduling engine. High-engagement tracks get scheduled more frequently at the hours when their listener segments are most active. Tracks that consistently cause tune-outs get flagged for rotation review. All processing happens on-premises — no data leaves your infrastructure.

Phase 8 Deliverables

• Analytics ring buffer in Mcaster1DNAS (extends existing song history ring buffer)
• Listener count timeseries API endpoint — per-mount, per-hour granularity
• Track engagement stats API — play count, listener delta per track change
• Analytics dashboard page in Mcaster1DNAS admin UI
• ICY 2.1 engagement signal receiver in DNAS (listener like/react events)
• Castit playlist intelligence module — query DNAS analytics API, score tracks
• Automatic rotation suggestion: flag overplayed / tune-out tracks in Castit UI
• Export: CSV / JSON analytics export from DNAS admin UI

Dependency Matrix

Git Repository Setup & Release Tags

git init
git remote add origin https://github.com/davestj/Mcaster1DSPEncoder.git
git add .
git commit -m "feat: Phase 5 — Podcast RSS, ICY 2.2 headers, Podcast + ICY 2.2 UI tabs"
git tag v0.2.0
git push -u origin main

Release Version Plan

Phase Summary