 karl
|
296a39a6c2
|
Isolate speech transport failures and keep experimental media usable
|
2026-10-13 21:09:43 +02:00 |
|
|
karl
|
f15ee07ed0
|
Preserve fractional-millisecond decoder evidence in timing evaluator
|
2026-10-13 20:52:35 +02:00 |
|
|
karl
|
2c3c91179b
|
Freeze normal-runtime speech timing regression without timestamp selection
|
2026-10-13 20:46:31 +02:00 |
|
|
karl
|
e68f4fcadc
|
Verify normal speech runtime and preserve independent quality limitations
|
2026-10-13 20:41:19 +02:00 |
|
|
karl
|
48f11dea3d
|
Add frozen shared-service speech retrieval and negative-control evaluator
|
2026-10-13 20:18:56 +02:00 |
|
|
karl
|
dab703413e
|
Add local spoken passages with Google retrieval and isolated runtime setup
|
2026-10-13 20:18:44 +02:00 |
|
|
karl
|
bc9459785d
|
Freeze independent English German speech and boundary validation inputs
|
2026-10-13 19:34:25 +02:00 |
|
|
karl
|
5f9df976eb
|
Document installed Windows 0.1.5 and complete Google speech timing improvement
|
2026-10-13 19:05:49 +02:00 |
|
|
karl
|
512ec4ea9f
|
Preserve exact frozen music and reranker manifest bytes across checkouts
|
2026-10-13 19:01:08 +02:00 |
|
|
karl
|
ec73d453e3
|
Record unseen music regressions and actual local reranker fidelity and cost
|
2026-10-13 18:59:45 +02:00 |
|
|
karl
|
cbf07b0575
|
Verify unseen music stress inputs against production decoding
|
2026-10-13 18:54:47 +02:00 |
|
|
karl
|
d27be302c8
|
Freeze unseen music stress corpus and independently verified gated DTW experiment
|
2026-10-13 18:51:50 +02:00 |
|
|
karl
|
b002399217
|
Retain bounded invalid ASR output and atomic failure evidence
|
2026-10-13 18:39:41 +02:00 |
|
|
karl
|
3ac1201909
|
Freeze independent DTW speech timing diagnostic and startup provenance
|
2026-10-13 18:33:21 +02:00 |
|
|
karl
|
705500425a
|
Freeze separate music metadata field diagnostic
|
2026-10-13 18:32:14 +02:00 |
|
|
karl
|
fede6ce3bb
|
Validate finer speech passages independently before Google ranking
|
2026-10-13 18:14:31 +02:00 |
|
|
karl
|
6353d02c04
|
Freeze actual Google music metadata and precise speech timing diagnostics
|
2026-10-13 18:11:07 +02:00 |
|
|
karl
|
8b4275d9b4
|
Prioritize media relevance improvements and refresh scale diagnostic lock
|
2026-10-13 18:06:22 +02:00 |
|
|
karl
|
0b6c958907
|
Preserve mixed evaluation evidence with reserved atomic report checkpoints
|
2026-10-13 17:30:52 +02:00 |
|
|
karl
|
d41c5b5e98
|
Recognize macOS extraction helper staging and verify release resources
|
2026-10-13 16:01:07 +02:00 |
|
|
karl
|
f1028e2689
|
Publish matching Windows CLI MCP package with document helper
|
2026-10-13 15:58:26 +02:00 |
|
|
karl
|
9ce60997db
|
Require document helper in Linux headless packages and verify Unix worker
|
2026-10-13 15:37:12 +02:00 |
|
|
karl
|
e250577740
|
Stage and require verified extraction helper in Windows release packages
|
2026-10-13 15:23:24 +02:00 |
|
|
karl
|
66bf7d7b47
|
Keep real model benchmark validity checks active under optimization
|
2026-10-13 13:54:33 +02:00 |
|
|
karl
|
d568bdb2b1
|
Measure genuine CPU query startup first input and warm latency
|
2026-10-13 13:53:03 +02:00 |
|
|
karl
|
59908adfa6
|
Validate gated Google retrieval and record unchanged localization quality
|
2026-10-13 13:49:23 +02:00 |
|
|
karl
|
af04cec27c
|
Measure complete CPU gated transcription and component costs
|
2026-10-13 13:47:23 +02:00 |
|
|
karl
|
f827df11c6
|
Verify VAD allocation limits and decoder overshoot contracts
|
2026-10-13 13:44:06 +02:00 |
|
|
karl
|
7a6ab9fc6a
|
Measure real CPU VAD on complete frozen media pool
|
2026-10-13 13:41:57 +02:00 |
|
|
karl
|
bc47e14971
|
Add bounded CPU VAD diagnostic with real model contract tests
|
2026-10-13 13:39:04 +02:00 |
|
|
karl
|
4a7c4acdcb
|
Prepare inspected static Metal runtime inventory for macOS staging
|
2026-10-13 13:26:16 +02:00 |
|
|
karl
|
07bbcb76e8
|
Reject stale native runtime build provenance before release staging
|
2026-10-13 13:04:58 +02:00 |
|
|
karl
|
baf59ba452
|
Preserve Linux archive failures and require current query-role runtime
|
2026-10-13 12:41:27 +02:00 |
|
|
karl
|
a8080f061f
|
Support reproducible XZ Linux archives without changing runtime contents
|
2026-10-13 12:31:17 +02:00 |
|
|
karl
|
193ab08f62
|
Evaluate frozen relevance hypothesis once on held-out Google CPU queries
|
2026-10-13 12:23:31 +02:00 |
|
|
karl
|
e7245fe15e
|
Freeze calibration-derived document and code relevance hypothesis
|
2026-10-13 12:17:37 +02:00 |
|
|
karl
|
ac02da724e
|
Record optimized WSL verification of GLib iterator backport
|
2026-10-13 12:05:46 +02:00 |
|
|
karl
|
3427abfd0f
|
Verify genuine Google CPU hybrid MCP from Windows archive
|
2026-10-13 11:59:33 +02:00 |
|
|
karl
|
3a336ca948
|
Verify extracted Windows CLI through actual MCP search and citations
|
2026-10-13 11:57:08 +02:00 |
|
|
karl
|
2be6b3d2b4
|
Support optional CLI and MCP in Windows source archives
|
2026-10-13 11:55:16 +02:00 |
|
|
karl
|
4e277a107f
|
Validate shared-engine lexical performance harness on Windows
|
2026-10-13 11:51:25 +02:00 |
|
|
karl
|
eabb1fbb54
|
Publish first usable Windows release instructions and validation
|
2026-10-13 11:20:23 +02:00 |
|
|
karl
|
5eeba29c0d
|
Validate audio stream timelines and add genuine Google transcript diagnostics
|
2026-10-13 10:39:31 +02:00 |
|
|
karl
|
8ef874c609
|
Backport official GLib iterator safety fix for Linux desktop
|
2026-10-13 10:35:50 +02:00 |
|
|
karl
|
43c43ebdc9
|
Add shared-engine calibration-only relevance diagnostics
|
2026-10-13 10:25:35 +02:00 |
|
|
karl
|
af6ce29125
|
Validate complete speech decode windows and retain rejected ASR output
|
2026-10-13 09:03:20 +02:00 |
|
|
karl
|
af7a40c183
|
Prepare bounded speech diagnostics and document mixed routing risks
|
2026-10-13 08:26:01 +02:00 |
|
|
karl
|
5d9cfc4ea0
|
Verify packaged workers from Unicode extraction directories
|
2026-10-13 08:13:33 +02:00 |
|
|
karl
|
aca0a0a4ce
|
Check packaged query and indexing contexts during CPU fallback
|
2026-10-13 08:00:16 +02:00 |
|
|
karl
|
88cd9c7f5a
|
Use measured query-only context across desktop CLI and MCP
|
2026-10-13 07:30:40 +02:00 |
|