Second voice UX iteration. Targets Marius's live-test pain points from today.
- **Voice-mode system prompt** (personality/VOICE_MODE.md, plumbed via
claude_session.build_system_prompt(voice_mode=True)) — when the voice
adapter starts a session, append voice-tailored instructions: short replies,
no markdown, no abbreviations, time without seconds, distances rounded
to "mii"/"milioane", no curly quotes / em-dash / ellipsis. Marius asked
for a "in-the-car friend" persona for voice.
- **Isolated voice session key** (router.py) — voice mode uses
`voice:<channel_id>` so it doesn't share context with the text adapter
on the same Discord channel. Fresh start, voice prompt applied
automatically without `/clear` ceremony. `/clear` drops both keys.
- **Metric units + Romanian thousands** (src/voice/normalize.py) —
`384.000 km` was being read as "trei sute optzeci și patru virgulă zero
zero zero km" because the dot was treated as decimal separator and `km`
wasn't expanded. New `normalize_thousands` collapses Romanian thousands
separators (`X.000`/`X.000.000`) before number expansion, and
`expand_units` handles km/kg/cm/mm/ml/ha/mp with correct Romanian
pluralization ("un kilometru", "două kilograme", "douăzeci de
centimetri", "o sută de kilometri" with "de" particle).
- **`/voice setvoice <M1-F5>` slash command** (discord_voice.py) — Discord
native autocomplete; swaps the live TTSQueue voice_id AND persists
voice.default_voice to config.json. No restart needed.
- **Verbal voice change** (src/voice/voice_commands.py — new module +
29 tests) — say "schimbă vocea pe M5" / "vorbește cu vocea F3" / "voce
em cinci" from inside the voice channel. Detector requires both a
trigger word (voce/vorbește/schimbă/treci pe) and a recognizable voice
ID (direct "M5", word form "em cinci", or fallback substring match for
Whisper-mangled forms like "unul cinci"=M5 and "Mâcinci"=M5). On
detection: live-swap, persist to config, mirror to chat with
`🎤 ... / 🔊 Voce → M5`, speak short ack in the NEW voice, skip
Claude. "pământinci" still can't be recovered (no recoverable digit
substring); user gets passthrough to Claude in that case.
- **Whisper initial_prompt** now lists the voice-command vocabulary so
STT biases toward producing clean "M5" / "F3" tokens instead of
inventing "pământ" / "unul" phonetic neighbors.
- **Fast barge-in** (pipeline.py EchoVoiceSink) — previously `ttsq.clear()`
only fired in `on_segment_done` (after 800ms silence + 2-3s STT ≈ 3s lag).
Now also fires from the sink as soon as VAD detects ≥2 consecutive
windows (~200ms) of sustained speech on Marius's user while Echo has
pending TTS frames. Single-window glitches don't cut Echo off; sustained
speech does. (Acoustic echo bleed-through still requires headphones —
no AEC in the bot.)
- Tests: 130 voice + router tests pass; updated test_router.py to expect
`/clear` to drop both text and voice session keys.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
End-to-end voice UX iteration after DAVE E2E shipped. Each change addresses a
real symptom Marius hit in live testing today:
- Kill the 3s filler ("mă gândesc"): Claude p50 is 4-7s so the filler always
fired BEFORE the response and collided with it. Removed all filler infra
from pipeline.py + tts_stream.py (FILLER_DELAY_S, _filler_task, push_filler,
load_thinking_wav, thinking.wav cache).
- Barge-in: ttsq.clear() at the top of on_segment_done drops stale frames so
a new utterance cuts off Echo's previous response cleanly.
- DTX silence flush: Discord stops sending RTP packets when the user goes
silent (DTX), so the inline silence-check in sink.write() never fired for
the trailing audio of an utterance — STT was missed entirely. Added a
background poller thread that checks the silence-flush condition every
200ms independent of incoming packets.
- Discord audio cadence fix: EchoStreamingAudioSource.read() blocked 100ms
per call when pcm_queue was empty, wrecking Discord's 20ms frame pacing →
client interpreted the stream as stutter and discarded leading frames
(Marius heard "4 de minute în București" instead of the full sentence).
Switched to get_frame_nowait() — instant return, silence frame on empty.
- RO time expansion: "23:09" was being read as "douăzeci și trei:nouă"
with literal colon. Added expand_time() with feminine-correct minute
formatting (un minut / două minute / douăzeci de minute / una de minute).
- Supertonic Unicode sanitize centralized in tools/tts.py: Romanian curly
quotes (`„`, `"`, `"`, `—`, `…`) crash Supertonic with HTTP 500. Map them
to ASCII at the synthesize() entry so BOTH voice mode and /audio command
are covered without duplication. normalize.py re-exports for compat.
- Whisper offline: WhisperModel(..., local_files_only=True) — no more
huggingface.co metadata GET on every startup. Model is already cached.
- Diagnostic logging across the chain: sink first-packet, VAD first-speech,
voice stream block (Claude → callback), push_text (text → clauses queued),
TTS pushed (clauses → frames). Lets future "spoke but Echo silent" bugs
pinpoint exactly where the chain breaks.
- Captured Supertonic curly-quote lesson in tasks/lessons.md.
All 76 voice tests pass.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Vendored fork: discord-ext-voice_recv 0.5.3a+echo.dave1
Patches the receive pipeline to handle Discord's mandatory DAVE E2E
encryption on voice gateway v=8. Without this, opus_decode raised
"corrupted stream" on every received packet in a DAVE-active room and
voice-to-voice never connected.
DAVE patch (vendor/discord-ext-voice-recv/reader.py):
- `_maybe_dave_decrypt(rtp_packet)`: gate mirrors discord.py 2.7.1
`voice_state.can_encrypt`. Uses davey's `can_passthrough(user_id)` to
branch — peers in passthrough send transport-only packets that pass
through verbatim; peers in DAVE epoch go through `davey.decrypt`.
- Hooked in `callback()` between transport decrypt and feed_rtp;
drops on decrypt failure without killing the reader thread.
- Bumps __version__ to '0.5.3a+echo.dave1' (PEP 440 local segment) so a
contract test can fail fast on accidental upstream-sync overwrite.
Pipeline fixes uncovered while testing DAVE end-to-end:
- src/voice/pipeline.py: silero-vad v6+ requires exactly 512 samples per
call at 16kHz; our 100ms window (1600 samples) was silently raising
ValueError → VAD always returned False → STT never fired. Slice the
window into 512-sample chunks. Bump whisper beam_size 1→5 and add a
Romanian `initial_prompt` — transcriptions go from "Eco salt." gibberish
to "Echo, salutare, te rog spune-mi cât este ora."
- src/voice/tts_stream.py: EchoStreamingAudioSource.read() returns a 20ms
silence frame instead of b'' on empty queue. Empty return is treated
by Discord as end-of-stream and kills the player, so any TTS pushed
later would be silently discarded.
- src/adapters/discord_voice.py: actually attach EchoStreamingAudioSource
to the voice client after the wakeup beep (chained via `after=`),
which was missing entirely — TTS frames had no consumer.
Tests:
- tests/test_voice_recv_dave.py: 11 unit + callback integration tests
covering bypass paths, can_passthrough gate, decrypt error handling.
- tests/test_voice_adapter_contract.py: +test_voice_recv_fork_version
and +test_voice_connection_state_has_dave_attrs guards against
upstream drift on either side.
Config:
- config.json: voice.allowed_user_ids whitelist for Marius's user id.
Status: voice-to-voice loop closes end-to-end (DAVE → VAD → Whisper →
Claude → Supertonic → audio out). Latency is ~8-13s per turn, which is
out of scope for this commit — see TODOS.md for the real-time UX
follow-up plan.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
src/voice/tts_stream.py (~280 lines):
- clause_segments(text, min_words=8): yield Romanian-aware clause chunks.
Split la punct (./!/?;:,) cu accumulation până min_words satisfied;
edge case text < min_words → single chunk. NU split mid-word/number/
currency. Romanian intonație de frază se rupe la sentence break — 8+
words minimizează seams.
- TTSQueue worker thread: text queue in → PCM frames out. Methods:
start/stop/push_text/push_filler/clear/is_empty. normalize_for_tts()
apply first, then clause_segments(), then Supertonic synth per chunk.
- EchoStreamingAudioSource(discord.AudioSource): read() pull from PCM
queue, 20ms frames (3840 bytes 48kHz s16le stereo). Eliminates RTP
gap between play() calls — single play() per session, source pulls.
- load_thinking_wav(): one-shot cache → 140 × 20ms frames (~2.8s)
pentru filler "Stai puțin să-mi adun gândurile".
- wav_to_pcm_20ms_frames(): WAV parser + ffmpeg subprocess resample
la 48kHz s16le stereo dacă nevoie.
Smoke test (în session): clause_segments behaviour OK, thinking.wav
loads, TTSQueue + EchoStreamingAudioSource construct clean. Integration
testing deferred la convergență (Pas 7 + Pas 11).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
personality/AGENTS.md — added ## Voice mode section after ## Platform
Formatting (logical fit: voice este încă un platform-specific register,
alături de Discord/WhatsApp formatting). 7 reguli aplicabile când
adapter_name == "discord-voice":
- 1-3 propoziții max
- fără markdown / fără bullet / fără linkuri
- numere/valute conversaționale ("treizeci de lei" nu "30 RON" —
normalize.py face conversia tehnică)
- lung/structurat → "L-am scris în chat." + text mirror
- ton ca la o cafea cu Marius, nu raport corporate
personality/TOOLS.md — added ### Discord Voice section după ### Whisper:
- ce e (bot ascultă/transcrie/răspunde rostit)
- "în voce" = /voice join, presence Listening, auto-leave 5min
- latency expectations ~5s perceived, filler peste 3s
- streaming TTS per clauză (zero gap)
- limitări (1-3 propoziții, STT pe cuvinte rare/acronime)
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Fix arhitectural general (beneficiu și pentru text adapters), nu doar voice.
src/claude_session.py:
- _session_locks: dict[str, threading.Lock] cu bootstrap lock pentru
lazy creation thread-safe.
- _get_session_lock(channel_id) helper.
- send_message() body wrapped în with _get_session_lock(channel_id).
- threading.Lock (NU asyncio.Lock) — send_message e sync subprocess.run
blocking; asyncio.Lock nu protejează cod sync rulat via to_thread.
- Per-channel granularity preserved — different channels run în paralel.
- send_message() public signature unchanged.
src/router.py:
- route_message(): dacă adapter_name == "discord-voice", prepend
[speaker:<user_name>] prefix (Config.get("voice.user_name", "user")).
- Original text variable left untouched for downstream paths.
- Text adapters: zero behavior change.
- route_message() public signature unchanged.
tests/test_claude_session_mutex.py — 6 tests REGRESSION-CRITICAL:
- same channel serializes (concurrent → mutex serializes, no overlap)
- same channel lock identity (same dict entry per channel_id)
- different channels run in parallel (overlap MUST fire)
- 3 channels all overlap
- contested acquire blocks then proceeds (policy: blocking, not fail-fast)
- lock released on subprocess exception (no deadlock on crash)
Acquisition policy: BLOCKING acquire bound by claude --timeout (5min default)
nu fail-fast — adapters already serialize via asyncio.to_thread queue, un
non-blocking acquire ar surface transient busy errors.
Test results: 82 passed (51 existing + 31 new). 2 PRE-EXISTING failures în
TestPromptInjectionProtection (stale assertion vs current prompt text) —
out of scope, recomand ticket separat.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pure functions pentru TTS text normalization (RO):
- strip_markdown: regex bold/italic/code/link/heading/list
- expand_numbers_ro: num2words pentru cardinals + decimal handling
("3.14" → "trei virgulă paisprezece", "3.05" → "trei virgulă zero
cinci" digit-by-digit la leading zero)
- expand_currency: formă naturală RO ("12.50 RON" → "doisprezece lei
și cincizeci de bani", "$25.99" → "douăzeci și cinci de dolari și
nouăzeci și nouă de cenți")
- expand_symbols: %/&/@/° + whitespace collapse
- expand_abbreviations: etc./dl./dna./nr./ş.a./ş.a.m.d.
- normalize_for_tts: full pipeline + hard truncate 200 cuvinte cu
"Restul l-am scris în chat."
Pipeline order: markdown → abbreviations → currency → numbers →
symbols → truncate. Currency BEFORE numbers — altfel "12.50 RON" se
degradează la "doisprezece virgulă cincizeci RON". Romanian "de"
particle rule: n>=20 AND (n%100 not in 1..19) → "o sută de lei",
"o sută cinci lei" (no "de"). n=1 with currency → "un dolar" /
"un leu" (article, nu cardinal).
35/35 tests pass: markdown(5), cardinals(6), decimals(4), currency
RON/USD/EUR/GBP mix(8), symbols(4), abbreviations(4), truncation(2),
edge cases empty/whitespace(2).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Parametrul `lang` era definit (DEFAULT_LANG = "ro") dar nu era inclus
in request-ul HTTP catre /v1/audio/speech. Adaugat "lang": lang in
body-ul JSON si lang="ro" explicit in _tts_synthesize().
OpenAPI-ul Supertonic confirma ca /v1/audio/speech accepta `lang`
ca parametru optional (OpenAISpeechRequest schema).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
newsletter-test și content-discovery eliminate la cererea lui Marius.
crontab check_newsletter_cercetasi.py de asemenea șters.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
email_digest.py folosea save_unread_emails() care salva în memory/kb/emails/.
Notițele KB trebuie create DOAR de heartbeat. Acum digest-ul face fetch
direct din IMAP (ca email_forward.py), fără side effects pe KB.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Instrucțiunea era prea restrictivă (doar formulare/documente "acționabile").
Acum include orice URL relevant: articole, linkuri de citit, resurse.
Același comportament adăugat și în HEARTBEAT pentru TL;DR din KB.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
La salvarea unui email forwardat, se extrage acum expeditorul original
din body și se elimină prefixul Fwd: din titlu — în loc de adresa lui Marius.
Corectat și fișierul deja salvat din 07 mai.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The idle-state action called /api/projects/approve, which 404'd because
idle workspace dirs have no approved-tasks.json entry to mutate. Now the
button opens the Propose modal pre-filled with the workspace slug so the
user actually has a path forward.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two structural fixes that together let users manage feature-branch
work without manual intervention:
Approval guard — `/plan/start` returns 409 `already_committed` if the
project status is approved/running/complete, unless the body opts in
with `force=true`. Frontend now renders "Re-planifică" instead of
"Planifică" on approved cards and gates it behind a confirm dialog
that threads `force=true` through. Prevents an accidental click from
wiping `status=approved` and burning a fresh planning subprocess.
Worktree awareness — projects can now declare that they target a
feature branch on an existing Gitea repo, not a repo-per-slug clone.
Three optional fields added to approved-tasks.json: `repo` (default
= slug), `branch` (feature branch to create), `base_branch` (default
main). Wired through `/p` flag parser in router.py, the dashboard
Propose modal's new "Avansat" section, and the night-execute prompt
which clones {repo} and creates {branch} from {base_branch} before
running ralph.
CLAUDE.md updated with both flows + the new schema fields.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
approved-tasks.json, dashboard/status.json, anaf-monitor/monitor.log
are auto-modified by background processes (heartbeat, cron jobs, ANAF
monitor). Untracking them stops the noisy "auto-commit from dashboard"
churn. Files stay on disk; readers (router._load_approved_tasks etc.)
already handle missing files by returning empty defaults.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
State files updated by dashboard/heartbeat/cron jobs, plus new KB
captures (samsung firmware todo, scout song reel, weekly youtube notes).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three fixes that together restore the planning UX:
- Dashboard reopen showed only a 500-char truncated excerpt of the last
assistant message. Backend now reads the Claude session JSONL directly
and returns full per-turn history; frontend iterates and renders all
bubbles, falling back to last_text_excerpt when the JSONL is missing.
- Phases never advanced because the agent ran /plan-* skills inline as
tool calls and the marker protocol was loose. Tightened the planning
prompt (mandatory PHASE_STATUS marker on the last line of every turn,
ban on inline phase invocation), and the frontend now auto-calls
/plan/advance when phase_ready=true.
- The phase strip never showed visual state because data-phase values
("office-hours") didn't match orchestrator phase names ("/office-hours").
Added normalizePhase + cleanup of PHASE_STATUS markers from rendered
bubbles.
Also bumps eco.py session-content truncation from 2k to 20k so /eco
session views aren't cut mid-response either.
Bumps last_text_excerpt fallback in planning_session.py from 500 to
50_000 so even when the JSONL is unavailable, the bubble isn't sliced
mid-word.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Hash fragments never reach the server so they're lost during login
redirects. ?file= survives the ?next= flow; #hash still works for
direct access when already logged in.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The reverse proxy strips /echo/ before Python, so next=/workspace.html.
Both the JS redirect and the server-side already-logged-in path now
prepend /echo to produce a valid public URL.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Pass current path as ?next= when bouncing unauthenticated requests
to /echo/login; after successful auth, JS reads and validates the
param (must start with /echo/, not /echo/login) before redirecting.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Fonturile inter-*.woff2 și tokens.css nu mai sunt referențiate —
Inter se încarcă din Google Fonts, tokens.css a fost înlocuit
de professional-theme.css în romfast-website (fișier greșit în repo).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Systemd user units get a minimal PATH that omits ~/.local/bin where
the claude binary lives, causing plan/respond to 500 on every call.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
_resolve_planning_key searches all active sessions by slug regardless of
adapter, so respond/finalize/cancel/advance work even when planning was
initiated from Discord or Telegram.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Tailscale Serve mapează /echo/* → 127.0.0.1:8088 (dashboard) și / →
:18789 (alt backend). Browser-ul calling /api/ralph/status (relative cu
absolute path la root domain) ajungea la 18789 care nu are endpoint Ralph
→ 502 Bad Gateway.
Fix: toate cele 6 URL-uri (5x fetch + 1x EventSource) folosesc acum prefix
/echo/api/ralph/* pentru a respecta routing-ul tailscale. Pattern consistent
cu workspace.html și index.html (verificat manual).
Endpoints atinse: /status, /<slug>/log, /<slug>/prd, /<slug>/stop,
/<slug>/rollback, /stream (SSE).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
approved-tasks.json mutat de start_planning_session cu status='planning'.
Sesiune activă: 14d2d96d-d4eb-4472-9b07-4a869909c564.
Confirmare empirică că flow-ul Discord/Telegram → modal/ForceReply →
PlanningOrchestrator funcționează end-to-end pe production.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
dashboard/api.py: adaug link "Ralph" (lucide bot icon) în NAV_HTML între
Workspace și KB. Pagina ralph.html se injectează corect cu nav-ul (verificat
live via curl pe :8088/ralph.html).
tests/test_e2e_planning_walkthrough.py (nou): 4 teste integration care
simulează scripted exact ce face un user pe Discord:
- click Planifică pe game-library cu UI scope → 4 faze (incl design-review)
- /office-hours → ceo → eng → design → final-plan.md stub scris pe disk
- "Dau drumul" → status approved + final_plan_path în approved-tasks.json
- description fără UI keywords → 3 faze (skip design)
- /cancel mid-planning → status revert pending, state cleared
- mesaj fără planning state → cade pe Claude main chat (NU orchestrator)
Subprocess `claude -p` mock-uit; testează tot wire-up-ul router → orchestrator
→ session și schema approved-tasks.json. Nu consumă credite.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replaces 5s polling on /echo/ralph.html with EventSource streaming and adds
a rollback control for the running Ralph cards.
Server (dashboard/handlers/ralph.py):
- /api/ralph/stream — Server-Sent Events. Emits `event: status` whenever a
signature over the projects' state changes (poll filesystem at 2s); emits
`event: heartbeat` every 30s to keep proxies happy. Disables proxy
buffering via X-Accel-Buffering:no.
- /api/ralph/<slug>/rollback (POST) — runs `git revert --no-edit HEAD` in
the project; falls back to `git reset --hard HEAD~1` only if revert
reports conflict. After rolling back the commit, decrements `passes` on
the last user story marked complete in prd.json (atomic temp+rename
write, same pattern as ralph_dag.py). Returns
`{success, message, reverted_commit, story_reverted, method}`.
- _ralph_validate_slug tightened to a strict regex (alphanum + dash +
underscore, ≤64 chars) plus explicit ../, /, \ rejection. All previously
accepted slugs still pass; URL-encoded traversal and shell metachars
now blocked before the filesystem is touched.
- _ralph_collect_status / _ralph_signature factored out of
handle_ralph_status so the SSE loop can reuse them and detect changes
cheaply.
Server (dashboard/api.py):
- HTTPServer → ThreadingHTTPServer with daemon_threads=True. SSE is a
long-lived response; without threading a single client would block all
other dashboard endpoints.
- /api/ralph/stream (GET) and /api/ralph/<slug>/rollback (POST) wired
into the dispatch.
Client (dashboard/ralph.html):
- EventSource('/api/ralph/stream') with permanent fallback to 5s polling
when readyState=CLOSED (no server, CORS blocked, browser without SSE).
- Indicator badge: 🟢 Live (SSE), ⏱ Polling (fallback), Offline.
- Rollback button (undo-2 icon) on running cards; native confirm() with
message: "Asta va da git revert HEAD pe <slug> și va decrementa ultima
story trecută. Continui?"
Tests (tests/test_dashboard_ralph_endpoint.py, +20 cases):
- Strict slug validator: underscore allowed, >64 rejected, special chars
/ backslash / URL-encoded traversal rejected.
- _ralph_collect_status + _ralph_signature: stable when nothing changes,
flips when project added or `passes` toggles.
- Rollback: invalid slug → 400, non-git project → 400, real two-commit
repo revert succeeds and decrements last passing story (US-002 goes
passes:false while US-001 stays passes:true), no-passing-stories case
succeeds with story_reverted=None, response shape contract, atomic
helper leaves no .tmp file behind.
- API routing smoke: confirms ThreadingHTTPServer + stream + rollback
references present in dashboard/api.py.
39/39 tests pass on tests/test_dashboard_ralph_endpoint.py. Pre-existing
failures in test_dashboard_constants.py::test_base_dir_is_echo_core (the
worktree dir is `echo-core-realtime`, not `echo-core`) and
test_dashboard_unified_index.py::test_index_has_all_panels are unrelated
to this change and reproduced on master.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
