voice-cat

Talon/voice-cat

Fork 0

Commit Graph

Author	SHA1	Message	Date
Talon	694494a5be	feat(M2): UDP voice/media plane -- SFU relay, Opus, AEAD, jitter buffer Adds the full voice pipeline: 14-byte binary frame header, ChaCha20-Poly1305 AEAD keyed from the TLS exporter, libopus encode/decode with FEC/PLC/DTX, an adaptive per-ssrc jitter buffer, a miniaudio capture/playback engine, an APM passthrough stub, and the UdpBinding/StreamAnnounce signaling chain wired through ConnSession/SessionRegistry into a new server-side SFU (MediaRelay) that decrypts and re-encrypts frames per channel member. Exit criterion verified: test_m2_voice — two headless clients relay 50 encrypted Opus frames through the server; ctest --preset m1-dev is 9/9 green. Also corrects protocol.md's UdpBinding diagram, which described the UDP-side binding packet as AEAD-sealed when it is in fact a plaintext bootstrap frame (separate from the TCP/TLS UdpBinding ack). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>	2026-06-16 01:31:14 +02:00
Talon	268d511f79	docs: initial design baseline for VoiceCat voice/text chat Establish the design spec in docs/ before implementation: - README: overview, locked decisions, principles, glossary - architecture: shared C++ core + C ABI, native UIs (Swift/C#), threading model, server design (SFU relay) - protocol: TCP/TLS control plane, protobuf Envelope + message catalog, connection lifecycle, extensibility rules - voice: UDP media frame format, per-channel Opus config, multi-stream model, two-sided noise reduction, VAD/PTT, jitter buffer, iOS ReplayKit screen-audio - security: mandatory encryption (TLS 1.3 + exported-key AEAD), TOFU server identity, admin-provisioned accounts, anti-replay - tech-stack: permissive-only deps (mbedTLS, libsodium, opus, miniaudio, webrtc-apm, ...), build tooling, no GPL/LGPL - deployment: zero-config self-host (Docker / binary / source) - roadmap: M0-M5 milestones, resolved decisions Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>	2026-06-15 20:47:09 +02:00

Author

SHA1

Message

Date

Talon

694494a5be

feat(M2): UDP voice/media plane -- SFU relay, Opus, AEAD, jitter buffer

Adds the full voice pipeline: 14-byte binary frame header, ChaCha20-Poly1305
AEAD keyed from the TLS exporter, libopus encode/decode with FEC/PLC/DTX,
an adaptive per-ssrc jitter buffer, a miniaudio capture/playback engine, an
APM passthrough stub, and the UdpBinding/StreamAnnounce signaling chain
wired through ConnSession/SessionRegistry into a new server-side SFU
(MediaRelay) that decrypts and re-encrypts frames per channel member.

Exit criterion verified: test_m2_voice — two headless clients relay 50
encrypted Opus frames through the server; ctest --preset m1-dev is 9/9
green. Also corrects protocol.md's UdpBinding diagram, which described the
UDP-side binding packet as AEAD-sealed when it is in fact a plaintext
bootstrap frame (separate from the TCP/TLS UdpBinding ack).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2026-06-16 01:31:14 +02:00

Talon

268d511f79

docs: initial design baseline for VoiceCat voice/text chat

Establish the design spec in docs/ before implementation:

- README: overview, locked decisions, principles, glossary
- architecture: shared C++ core + C ABI, native UIs (Swift/C#),
  threading model, server design (SFU relay)
- protocol: TCP/TLS control plane, protobuf Envelope + message
  catalog, connection lifecycle, extensibility rules
- voice: UDP media frame format, per-channel Opus config,
  multi-stream model, two-sided noise reduction, VAD/PTT,
  jitter buffer, iOS ReplayKit screen-audio
- security: mandatory encryption (TLS 1.3 + exported-key AEAD),
  TOFU server identity, admin-provisioned accounts, anti-replay
- tech-stack: permissive-only deps (mbedTLS, libsodium, opus,
  miniaudio, webrtc-apm, ...), build tooling, no GPL/LGPL
- deployment: zero-config self-host (Docker / binary / source)
- roadmap: M0-M5 milestones, resolved decisions

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

2026-06-15 20:47:09 +02:00

2 Commits