voice_audio_multi/tests/fun_asr_nano/frontend_reference.json
Fabio c36510c21e
Some checks failed
Linux Build / Linux cpu (push) Has been cancelled
Linux Build / Linux cpu-clang (push) Has been cancelled
Linux Build / Linux vulkan (push) Has been cancelled
mac-build / macOS CPU (push) Has been cancelled
Nix Build / Nix (cpu) (push) Has been cancelled
Nix Build / Nix (vulkan) (push) Has been cancelled
windows-build / Windows CPU (push) Has been cancelled
Initial import: audio.cpp with Orange Pi documentation
2026-11-04 19:05:22 +01:00

357 lines
9.5 KiB
JSON

{
"schema_version": 2,
"reference": "transformers.FunAsrNanoFeatureExtractor",
"transformers_commit": "48e7f65fb274172e15aa88875d780c67c37606c7",
"feature_extractor_sha256": "cb0e31712b264a62e0aa10f7ff23ef531bca02dbee1f378b8947efbdf3b85ee2",
"torch_version": "2.11.0+cu130",
"torchaudio_version": "2.11.0+cu130",
"sample_wav": "sample_16k.wav",
"sample_wav_sha256": "a8096a0c2ac136bcaf70143dc80ba576ade8dfbf7f9f30ecdec29b920ab3e3cb",
"data_file": "frontend_reference.bin",
"data_format": "little-endian-float32",
"data_sha256": "12a42c35f0ca863e5ff17c56394468ba91cb2fe9a4de67a59001a77cf9e0363e",
"sample_rate": 16000,
"feature_size": 80,
"frame_length_ms": 25,
"frame_shift_ms": 10,
"lfr_m": 7,
"lfr_n": 6,
"fixtures": {
"silence": {
"samples": 16000,
"frames": 17,
"width": 560,
"min": -15.942384719848633,
"max": -15.942384719848633,
"mean": -15.942384719848633,
"l2": 1555.5062464965847,
"first32": [
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633
],
"last32": [
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633
],
"waveform": {
"offset_f32": 0,
"count": 16000
},
"features": {
"offset_f32": 16000,
"count": 9520
}
},
"impulse": {
"samples": 16000,
"frames": 17,
"width": 560,
"min": -15.942384719848633,
"max": 3.2924442291259766,
"mean": -15.599400718495673,
"l2": 1536.8208434990697,
"first32": [
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633
],
"last32": [
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633,
-15.942384719848633
],
"waveform": {
"offset_f32": 25520,
"count": 16000
},
"features": {
"offset_f32": 41520,
"count": 9520
}
},
"sine_440hz": {
"samples": 16000,
"frames": 17,
"width": 560,
"min": -10.94229793548584,
"max": 4.429123401641846,
"mean": -7.612086516202373,
"l2": 796.5784973954068,
"first32": [
-8.902217864990234,
-7.893123626708984,
-7.683259963989258,
-8.373307228088379,
-9.063929557800293,
-8.556164741516113,
-7.122377395629883,
-6.354111194610596,
-6.597945213317871,
-9.190327644348145,
-8.430506706237793,
-5.84726619720459,
0.3718216121196747,
3.6082963943481445,
4.427730560302734,
3.3434324264526367,
-0.4013015031814575,
-6.9655890464782715,
-5.87312650680542,
-4.906649589538574,
-5.516244411468506,
-6.746131896972656,
-5.554996013641357,
-5.721648693084717,
-7.326694011688232,
-5.929301738739014,
-6.312660217285156,
-7.367210865020752,
-6.188997745513916,
-7.128389358520508,
-6.936490058898926,
-6.658333778381348
],
"last32": [
-10.55267333984375,
-10.470776557922363,
-10.453227043151855,
-10.47970962524414,
-10.690077781677246,
-10.710362434387207,
-10.666672706604004,
-10.772344589233398,
-10.792530059814453,
-10.789617538452148,
-10.870523452758789,
-10.887447357177734,
-10.884880065917969,
-10.852453231811523,
-10.761302947998047,
-10.912212371826172,
-10.925711631774902,
-10.91910171508789,
-10.913459777832031,
-10.814454078674316,
-10.699851036071777,
-10.876879692077637,
-10.893570899963379,
-10.742762565612793,
-10.122283935546875,
-10.580676078796387,
-10.839366912841797,
-10.707252502441406,
-10.054222106933594,
-10.470293998718262,
-10.615829467773438,
-10.527753829956055
],
"waveform": {
"offset_f32": 51040,
"count": 16000
},
"features": {
"offset_f32": 67040,
"count": 9520
}
},
"sample_16k": {
"samples": 225151,
"frames": 235,
"width": 560,
"min": -15.942384719848633,
"max": 6.347103118896484,
"mean": -5.632492880177623,
"l2": 2549.3456314076384,
"first32": [
-8.025833129882812,
-7.580291748046875,
-8.279046058654785,
-8.900419235229492,
-9.515005111694336,
-10.775506019592285,
-12.052884101867676,
-11.757328987121582,
-13.11245346069336,
-15.366243362426758,
-14.615835189819336,
-13.44446849822998,
-12.274049758911133,
-12.53348159790039,
-13.522049903869629,
-13.94956111907959,
-13.105130195617676,
-12.761933326721191,
-13.164908409118652,
-13.763184547424316,
-14.2503023147583,
-12.552626609802246,
-12.877581596374512,
-12.61111068725586,
-12.65695858001709,
-13.439993858337402,
-12.562033653259277,
-12.736815452575684,
-13.652838706970215,
-14.115615844726562,
-12.287749290466309,
-11.618743896484375
],
"last32": [
-10.266275405883789,
-9.827715873718262,
-9.91140365600586,
-10.755105018615723,
-11.691581726074219,
-11.801953315734863,
-11.057150840759277,
-11.519274711608887,
-11.85002613067627,
-11.05031967163086,
-10.317023277282715,
-10.019948959350586,
-10.43407154083252,
-10.095555305480957,
-9.02872085571289,
-10.206125259399414,
-9.60042667388916,
-10.06050968170166,
-9.961405754089355,
-10.69766902923584,
-9.176285743713379,
-9.457172393798828,
-9.47586441040039,
-9.176664352416992,
-9.69524097442627,
-9.745139122009277,
-9.755215644836426,
-10.710921287536621,
-10.568866729736328,
-10.585000991821289,
-11.155245780944824,
-11.610282897949219
],
"features": {
"offset_f32": 76560,
"count": 131600
}
}
}
}