Expose transcribe.cpp Q8 at a Handy-compatible root filename

.gitattributes

@@ -44,3 +44,4 @@ report/assets/unseen/unseen-english-domains.png filter=lfs diff=lfs merge=lfs -t
 report/assets/unseen/unseen-eurospeech.png filter=lfs diff=lfs merge=lfs -text
 report/assets/unseen/unseen-language-coverage.png filter=lfs diff=lfs merge=lfs -text
 transcribe-cpp/orukeet-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text
+orukeet-transcribe-cpp-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text

ARTIFACTS.json

@@ -51,7 +51,7 @@
     },
     "README.md": {
       "bytes": 13115,
-      "sha256": "c8be4e891a2cb526dd2f8e24e02a001ed8f4b4ac2111674b2a9242a65b08da07"
+      "sha256": "e81cc20f6218f878aeaf496facb727971702abda1e06bb9e9ecbb5f60c73c275"
     },
     "CITATION.bib": {
       "bytes": 562,
@@ -614,16 +614,16 @@
       "sha256": "29d50f34b1ffe4b11665aaeefdaf63a22796c53c38234cd9d6f06c5153c9e13a"
     },
     "transcribe-cpp/README.md": {
-      "bytes": 3091,
-      "sha256": "15598b8d00955d4b9c657b8034d02143c7981d2e254853c1db9cb1e95fdb620d"
+      "bytes": 3281,
+      "sha256": "76385d17cbe11d605aa02db6c19ef98aa6737d744e9dbae39d0c667cc47a9dc6"
     },
     "transcribe-cpp/validation.json": {
       "bytes": 5409,
       "sha256": "0bca4eabd62c784a14eb486aeead27868b8014d98b62378012b75669e19fff42"
     },
     "transcribe-cpp/manifest.json": {
-      "bytes": 939,
-      "sha256": "11e14b26e08b06710ff739d12ebd3eb98d91d6ce859ecc1f559a092a63357a6c"
+      "bytes": 1015,
+      "sha256": "464713d1b57e82f63deedbca10cecbd39f6be5d7718f0bf4ae5a6b5a429ac668"
     },
     "transcribe-cpp/convert-orukeet.patch": {
       "bytes": 1635,
@@ -632,6 +632,10 @@
     "transcribe-cpp/orukeet-Q8_0.gguf": {
       "bytes": 739508608,
       "sha256": "cad2f52ac91cad829279422301989687c2cf02e19157352ed25ea501b90dbb7e"
+    },
+    "orukeet-transcribe-cpp-Q8_0.gguf": {
+      "bytes": 739508608,
+      "sha256": "cad2f52ac91cad829279422301989687c2cf02e19157352ed25ea501b90dbb7e"
     }
   }
 }

README.md

@@ -149,7 +149,7 @@ times; it does not return emotion, speaking-style or speaker-diarization scores.
 
 ## transcribe.cpp and Handy-compatible GGUF
 
-[`transcribe-cpp/orukeet-Q8_0.gguf`](transcribe-cpp/orukeet-Q8_0.gguf) is a Q8 export of the same r3 checkpoint for [transcribe.cpp](https://github.com/cjpais/transcribe.cpp). It uses the existing `parakeet` architecture and requires no Gabor-specific runtime. CPU and Apple Metal checks use the exact `transcribe-cpp` 0.2.0 dependency pinned by Handy.
+[`orukeet-transcribe-cpp-Q8_0.gguf`](orukeet-transcribe-cpp-Q8_0.gguf) is a Q8 export of the same r3 checkpoint for [transcribe.cpp](https://github.com/cjpais/transcribe.cpp). It uses the existing `parakeet` architecture and requires no Gabor-specific runtime. CPU and Apple Metal checks use the exact `transcribe-cpp` 0.2.0 dependency pinned by Handy.
 
 This file has a different tensor layout from the native NeMo-Speech.cpp GGUFs above. Select the export for your runtime. [Conversion, checksums and validation](transcribe-cpp/README.md).
 
@@ -159,7 +159,7 @@ This file has a different tensor layout from the native NeMo-Speech.cpp GGUFs ab
 |:--|:--|--:|
 | NeMo source | `orukeet-v0.1.0.nemo` | 2,509,342,720 |
 | Native Q8 | `orukeet-v0.1.0-q8.gguf` | 714,456,704 |
-| transcribe.cpp Q8 | `transcribe-cpp/orukeet-Q8_0.gguf` | 739,508,608 |
+| transcribe.cpp Q8 | `orukeet-transcribe-cpp-Q8_0.gguf` | 739,508,608 |
 | Native F16 | `orukeet-v0.1.0-f16.gguf` | 1,296,681,088 |
 | ONNX INT8 archive | `onnx/sherpa-onnx-orukeet-v0.1.0-int8.tar.bz2` | 486,807,585 |
 

orukeet-transcribe-cpp-Q8_0.gguf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cad2f52ac91cad829279422301989687c2cf02e19157352ed25ea501b90dbb7e
+size 739508608

transcribe-cpp/README.md

@@ -2,17 +2,19 @@
 
 This Q8_0 export uses transcribe.cpp's existing Parakeet TDT v3 implementation. The 12,288 fitted Gabor kernels are materialized as ordinary convolution weights. There are no additional runtime operators or dependencies.
 
-- File: `transcribe-cpp/orukeet-Q8_0.gguf` (739,508,608 bytes)
+- File: `orukeet-transcribe-cpp-Q8_0.gguf` (739,508,608 bytes)
 - SHA-256: `cad2f52ac91cad829279422301989687c2cf02e19157352ed25ea501b90dbb7e`
 - Source: Orukeet r3, SHA-256 `031c8ddab4845aeced904a7cde8e8aa57993b2e344716cf83a545b079c473b56`
 - Weights: [CC BY-SA 4.0](../LICENSE-WEIGHTS). Orukeet is an adaptation of NVIDIA Parakeet TDT v3; retain [attribution](../NOTICE.md).
 
+The root-level filename follows Handy's existing cache and delete layout. The earlier `transcribe-cpp/orukeet-Q8_0.gguf` path remains available and contains identical bytes.
+
 ## Run
 
 Build [transcribe.cpp](https://github.com/cjpais/transcribe.cpp/tree/585b98f7e66777d16f2da734ceedaa7398060fa7) normally, then run:
 
 ```sh
-./build/bin/transcribe-cli --model orukeet-Q8_0.gguf recording.wav
+./build/bin/transcribe-cli --model orukeet-transcribe-cpp-Q8_0.gguf recording.wav
 ```
 
 Use 16 kHz mono audio. This is offline transcription in 25 languages, with language detection and token timestamps. Translation and streaming recognition are not supported. Use this layout with transcribe.cpp; the root-level `orukeet-v0.1.0-q8.gguf` is for NeMo-Speech.cpp.

transcribe-cpp/manifest.json

@@ -9,12 +9,15 @@
     "sha256": "031c8ddab4845aeced904a7cde8e8aa57993b2e344716cf83a545b079c473b56"
   },
   "export": {
-    "filename": "transcribe-cpp/orukeet-Q8_0.gguf",
+    "filename": "orukeet-transcribe-cpp-Q8_0.gguf",
     "bytes": 739508608,
     "sha256": "cad2f52ac91cad829279422301989687c2cf02e19157352ed25ea501b90dbb7e",
     "architecture": "parakeet",
     "tensor_count": 697,
-    "quantization": "Q8_0"
+    "quantization": "Q8_0",
+    "compatible_aliases": [
+      "transcribe-cpp/orukeet-Q8_0.gguf"
+    ]
   },
   "converter": {
     "repo": "https://github.com/cjpais/transcribe.cpp",