Default temperature 1.0

README.md

@@ -36,7 +36,7 @@ training data: [datasets/machinekind/polish-synthetic-speech-voxcpm2](https://hu
 | v1 model with the new voice slot, before fine-tuning | 2.91 | 2.7 % | 0.725 |
 
 WER counts Whisper's spelling too: the teacher's two "errors" are "krutki" for *krótki* and "HWSR" for the robot's name.
-Speaking rate: the released default **tempo is 0.8** (`length_scale`); at 1.0 v2 speaks about 19 % slower than the teacher.
+Defaults: **temperature 1.0**, **tempo 0.8** (`length_scale`; at 1.0 v2 speaks about 19 % slower than the teacher), neutral style 7.
 
 ## Files
 
@@ -65,7 +65,7 @@ ids = np.array(intersperse(text_to_sequence(text, ["polish_mixed_cleaners"])[0],
 vj = json.load(open("onnx/voices.json"))
 spk = np.array(vj["speakers"]["0"]["emb"], np.float32) + np.array(vj["styles"]["7"]["emb"], np.float32)   # 7 = neutral style
 sess = ort.InferenceSession("onnx/matcha_pl_t4.onnx")
-wav, n = sess.run(None, {"x": ids[None], "x_lengths": np.array([ids.size]), "scales": np.array([0.8, 0.8], np.float32), "spk_emb": spk[None]})
+wav, n = sess.run(None, {"x": ids[None], "x_lengths": np.array([ids.size]), "scales": np.array([1.0, 0.8], np.float32), "spk_emb": spk[None]})
 sf.write("out.wav", wav[0, :n[0]], 22050)   # scales = [temperature, length_scale (tempo)]
 ```
 

english/cleaner.js

@@ -1,7 +1,7 @@
 // Browser port of matcha_patch/polish_cleaners.py + polish_mixed_cleaners: Polish text normalisation, espeak-ng (WASM) IPA,
 // English words in {braces} phonemized with the en-us voice, symbol ids. Must match the Python cleaners character for character.
 import { initialize, setVoice, getPhonemes } from "https://cdn.jsdelivr.net/npm/espeak-phonemizer@0.1.2/src/espeak.mjs";
-import { SYMBOL_TO_ID } from "./symbols.js?v=20261005a";
+import { SYMBOL_TO_ID } from "./symbols.js?v=20261005b";
 
 const PL_WORD = "0-9a-ząćęłńóśźż";
 const esc = (s) => s.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");

english/english.js

@@ -3,7 +3,7 @@
 // unlisted words; a frequency table is too big for the browser). Inflected forms whose full form is a common Polish word
 // (wina, legendy, deskę) are precomputed in english_data.json -> pl_inflected_block.
 let D = null;
-export async function initEnglish(url = new URL("./english_data.json?v=20261005a", import.meta.url)) {
+export async function initEnglish(url = new URL("./english_data.json?v=20261005b", import.meta.url)) {
   if (!D) {
     const j = await (await fetch(url)).json();
     D = { cur: new Set(j.curated), rap: new Set(j.read_as_polish), suf: j.suffixes, loan: j.loan_stems, block: new Set(j.pl_inflected_block) };

onnx/voices.json

@@ -1 +1 @@
-{"default": "male_voxcpm_1", "targets": [], "presets": {"male_voxcpm_1": {"label": "male_voxcpm_1 (synthetic voice)", "weights": {"0": 1.0}}}, "speakers": {"0": {"name": "male_voxcpm_1", "emb": [-0.171126, 0.322866, -0.493196, -0.251865, -1.496564, -0.773552, 0.672265, -0.925804, -0.569764, -0.062601, 0.368567, -0.555231, -0.010228, 0.525713, 1.064089, 1.084695, -0.108082, -1.248569, 0.401326, -0.809678, 0.146417, 0.637237, 0.170811, -0.746055, 0.30452, -0.758564, -0.460776, -0.647524, -0.687801, 2.077753, -0.159227, -0.075347, 0.608661, 0.566903, -0.023465, -1.022551, -0.390619, -0.467591, 0.456134, 0.381035, -0.043261, -0.79233, -1.292068, -0.338539, 0.090066, -0.007928, 1.320292, -0.026984, -0.945175, -0.495162, -0.239323, 0.335876, -0.208299, 1.380403, -0.118773, -1.187929, 1.033132, 0.574047, 0.126154, 0.615193, -0.638149, -0.652656, 0.079094, 1.707609]}}, "styles": {"0": {"label": "flat-range · slow", "emb": [-0.145681, 0.028514, 0.031073, 0.213126, 0.141315, -0.091171, 0.020456, -0.050942, 0.116065, -0.037322, -0.050199, -0.128892, -0.080357, 0.045101, 0.012145, -0.013334, 0.032367, 0.00523, 0.016566, -0.002529, 0.094458, 0.054667, -7e-06, -0.011532, -0.006527, -0.130739, -0.04346, -0.025447, -0.056555, 0.02478, 0.035768, -0.007505, -0.064534, 0.114622, -0.005323, -0.011425, -0.081168, -0.026979, -0.054348, 0.063037, 0.021073, 0.113498, 0.125257, 0.107236, -0.032194, 0.021723, 0.069703, 0.023842, 0.102836, -0.123746, 0.012705, -0.015574, -0.023926, 0.049007, 0.077019, 0.142015, -0.022084, -0.036438, 0.126887, -0.074276, -0.087569, -0.009835, -0.031884, 0.1038]}, "1": {"label": "flat-range · slow · question", "emb": [-0.065793, 0.16722, -0.040421, 0.10364, 0.149176, -0.021905, -0.0432, -0.050239, -0.040979, -0.090244, -0.097514, -0.184097, -0.077265, 0.04736, -0.055627, -0.118585, 0.066956, -0.061624, 0.048341, 0.052993, 0.177459, 0.01611, -0.109275, -0.070324, 0.041396, 0.006175, 0.020192, -0.063335, 0.00341, -0.05934, -0.201707, 0.057273, 0.006712, 0.016891, 0.020629, -0.063877, -0.011998, -0.184993, -0.020477, 0.005055, -0.082237, 0.120278, -0.023095, 0.012626, 0.12684, 0.096845, -0.126149, 0.069004, -0.104394, 0.008861, 0.031055, -0.069648, 0.03642, 0.04356, 0.024144, 0.107632, 0.092163, 0.005802, 0.17045, -0.057356, -0.10544, 0.03624, -0.143395, 0.086159]}, "2": {"label": "flat-range · normal", "emb": [0.015198, 0.005387, -0.070077, 0.085694, 0.052682, -0.069297, -0.052193, -0.011528, 0.103583, 0.036779, -0.018627, -0.018627, -0.054331, -0.057561, 0.073174, 0.013283, 0.107597, 0.027577, 0.007558, 0.048043, -0.014568, 0.021438, -0.040362, -0.023572, 0.020275, -0.050076, -0.064918, 0.025708, -0.020152, -0.007477, 0.011547, 0.020923, -0.030765, -0.026945, -0.108283, 0.071716, -0.079583, -0.020836, 0.00641, 0.041376, -0.011752, 0.124817, 0.074644, 0.055393, 0.074216, 0.001747, 0.055778, -0.037595, 0.018188, -0.117185, -0.054989, 0.057521, 0.038923, 3.1e-05, -0.018487, 0.037477, 0.0652, 0.003342, 0.0622, 0.02653, 0.015227, -0.062551, 0.058737, -0.038891]}, "3": {"label": "flat-range · normal · question", "emb": [-0.033626, 0.123205, -0.104265, 0.116839, 0.158891, 0.03783, -0.124739, -0.13248, 0.047867, 0.023742, 0.102209, -0.103134, -0.015378, 0.050243, 0.184138, -0.05144, 0.137179, 0.063629, 0.10753, -0.026578, -0.051683, 0.090953, 0.096812, -0.112768, 0.075321, -0.093111, 0.098713, -0.018047, 0.111272, -0.088065, -0.118353, 0.12843, -0.041105, -0.091738, -0.013021, 0.096103, -0.080562, -0.139384, -0.049917, -0.024333, -0.139616, 0.064256, 0.010544, -0.032663, -0.035371, 0.009209, -0.197999, 0.041773, -0.06381, -0.09734, 0.00059, 0.013481, 0.011969, -0.06911, 0.026348, -0.020739, 0.100743, 0.028176, 0.141968, 0.027852, -0.026274, 0.098182, -0.024583, 0.015864]}, "4": {"label": "flat-range · fast", "emb": [0.079858, -0.032991, -0.141646, -0.0183, -0.05767, -0.010497, -0.130122, -0.063902, 0.006765, 0.108658, 0.042958, -0.02097, 0.050789, -0.10057, 0.101453, 0.113868, 0.042049, 0.099299, -0.084551, -0.002367, -0.086793, 0.08406, -0.099978, 0.061491, 0.097875, 0.108743, -0.111331, 0.084997, 0.144157, 0.00984, 0.065044, 0.041029, 0.064102, -0.095855, -0.145943, 0.078539, 0.045341, 0.036196, 0.068478, -0.028343, -0.042336, 0.131945, 0.046076, -0.006211, 0.078042, -0.122365, 0.007158, -0.175239, -0.052579, -0.182534, -0.110177, 0.126633, 0.037948, 0.037947, -0.06254, -0.004245, 0.158694, 0.066889, -0.018682, -0.003524, 0.125085, 0.024064, 0.130969, -0.106359]}, "5": {"label": "flat-range · fast · question", "emb": [0.09611, 0.034653, -0.220507, -0.148164, 0.024787, -0.047649, -0.211915, 0.173699, -0.072696, 0.068309, 0.094034, 0.072822, 0.019759, -0.108546, 0.158991, -0.032414, -0.033625, 0.035397, -0.038269, 0.020347, -0.026651, 0.06203, -0.057607, -0.056806, 0.105391, 0.188259, 0.159449, 0.088129, 0.173849, 0.078067, -0.155068, 0.117547, 0.10826, -0.139548, -0.019679, 0.050196, -0.005208, -0.135763, -0.048872, -0.156271, -0.068358, 0.079436, -0.067806, -0.098738, -0.000494, -0.008177, -0.207378, -0.158861, -0.088763, -0.091702, -0.054759, 0.171019, 0.123302, -0.038804, -0.038359, -0.009134, 0.188344, 0.213565, 0.065066, 0.131621, 0.000913, 0.027884, 0.028019, -0.047102]}, "6": {"label": "mid-range · slow", "emb": [-0.108414, -0.003766, 0.026759, 0.090972, 0.10488, 0.021303, 0.077811, 0.025143, 0.086748, -0.043222, -0.055501, -0.021596, -0.058954, 0.001334, 0.023127, -0.064402, -0.036275, -0.027464, -0.026106, -0.042851, 0.060475, -0.092045, 0.021875, 0.021948, -0.03574, -0.059152, -0.073984, 0.022593, -0.105647, 0.0545, 0.035674, -0.054579, -0.073088, 0.100104, 0.035317, -0.017922, -0.046948, -0.142279, -0.01218, 0.065619, 0.045547, -0.026938, 0.060306, 0.052985, 0.078487, 0.10227, 0.052383, 0.094831, 0.118608, 0.057531, 0.049244, -0.069276, 0.011932, -0.007232, 0.115472, 0.066428, -0.100617, -0.001337, 0.06223, -0.00322, -0.097148, -0.034642, -0.11133, 0.029218]}, "7": {"label": "mid-range · slow · question", "emb": [-0.029557, 0.133307, 0.098692, 0.122858, 0.09645, 0.008557, -0.014066, -0.135018, -0.079446, -0.12148, -0.135168, -0.129205, -0.064973, 0.125791, -0.074336, 0.01901, -0.015607, -0.046001, 0.083561, -0.01328, 0.075397, -0.012248, 0.024806, -0.042582, -0.084804, -0.042625, 0.019341, -0.129714, 0.029283, 0.029565, -0.080985, 0.070864, -0.059116, 0.07597, 0.106906, -0.044583, -0.017669, -0.20186, -0.141748, 0.010677, 0.008098, -0.023605, 0.056758, 0.098303, -0.034178, 0.045052, -0.183866, 0.117161, -0.002713, -0.024853, 0.015352, -0.082478, -0.014168, 0.110169, -0.000971, -0.01839, 0.054924, -0.169824, 0.123016, -0.045664, -0.173564, 0.084706, -0.13797, -0.031571]}, "8": {"label": "mid-range · normal", "emb": [0.047679, -0.033757, 0.012959, -0.020053, -0.017468, -0.00887, 0.010025, 0.062036, 0.099756, 0.013468, -0.017102, -0.037922, 0.03251, 0.001108, 0.008804, 0.02545, -9e-05, -0.001319, -0.010407, 0.004526, -0.03449, -0.04883, 0.001763, 0.010887, -0.014453, 0.03218, -0.055419, 0.059566, -0.003377, 0.022355, 0.007696, -0.022474, 0.013298, 0.010364, -0.023911, 0.041498, -0.037128, 0.003917, 0.013321, -0.010775, -0.001142, -0.059385, 0.010826, 0.003236, 0.048636, 0.028829, 0.040092, 0.028113, 0.009203, -0.027532, 0.003704, 0.009829, 0.002783, -0.042211, 0.032808, 0.036177, -0.001528, 0.017659, -0.00515, 0.016582, 0.046999, -0.007643, 0.016614, -0.050302]}, "9": {"label": "mid-range · normal · question", "emb": [0.100659, -0.035835, -0.122456, -0.055806, -0.030154, 0.026734, -0.065558, -0.071907, -0.035, -0.033529, -0.074552, -0.03608, -0.054054, -0.077936, -0.061722, -0.055407, 0.160748, 0.000847, -0.061431, 0.010338, 0.100741, 0.014006, -0.102683, -0.072087, 0.03347, 0.006653, 0.183114, -0.049406, -0.037271, 0.041016, -0.186704, -0.042957, 0.027818, 0.038135, 0.052061, 0.089554, 0.044856, -0.021578, -0.071574, 0.037154, -0.032053, -0.05052, -0.023911, -0.050673, -0.013418, 0.040641, -0.213389, 0.059986, -0.13572, 0.073084, -0.129966, -0.05335, 0.188839, 0.034079, -0.022861, -0.074348, 0.077725, -0.061973, -0.01645, 0.041444, -0.007721, 0.011829, -0.024506, 0.00019]}, "10": {"label": "mid-range · fast", "emb": [0.148559, -0.117011, -0.108052, -0.079113, -0.151191, -0.009634, -0.05918, 0.006944, 0.0017, 0.115003, 0.033147, 0.08128, 0.118907, -0.075013, -0.014178, 0.08572, -0.014366, 0.092867, -0.012872, -0.028587, -0.077081, 0.010422, -0.063002, 0.000577, -0.009214, 0.135065, -0.056282, 0.013967, 0.067966, -0.058367, 0.023978, 0.070172, 0.075121, -0.092142, -0.06845, 0.120971, 0.014942, 0.059432, 0.032089, -0.000947, -0.023865, -0.014222, 0.013042, -0.099835, -0.027346, -0.049702, 0.015537, -0.066083, -0.075356, -0.094995, -0.060072, 0.054041, 0.036734, -0.039968, -0.078773, -0.020267, -0.01236, 0.055531, -0.081482, 0.057737, 0.08034, -0.001431, 0.14718, -0.100518]}, "11": {"label": "mid-range · fast · question", "emb": [0.073619, 0.043859, -0.05577, -0.034364, -0.257909, 0.070455, -0.16292, -0.010065, -0.044395, 0.102808, -0.050129, 0.0522, 0.193897, -0.029656, 0.007619, 0.052111, 0.04524, 0.049192, 0.113122, 0.129318, -0.029497, 0.10029, -0.2246, 0.008363, 0.124819, 0.078415, 0.031787, -0.101723, 0.063401, -0.14017, -0.119636, 0.044516, 0.090903, -0.005523, -0.089667, 0.139016, 0.156393, 0.024767, -0.054849, -0.176828, -0.139236, -0.009136, -0.24576, -0.182452, -0.160701, -0.003506, -0.06909, -0.035963, -0.067486, -0.004653, 0.015554, 0.119868, 0.013078, -0.071847, -0.064107, -0.047181, 0.090308, 0.078807, 0.023879, 0.118498, 0.09614, 0.184368, -0.05321, -0.100499]}, "12": {"label": "wide-range · slow", "emb": [-0.11099, 0.018829, 0.151204, 0.0139, 0.021509, 0.077328, 0.178788, -0.00928, 0.037616, -0.079435, -0.026652, 0.004519, -0.03194, -0.02984, -0.083479, 0.028554, -0.129474, -0.073987, 0.002037, -0.035017, 0.00193, -0.117057, 0.072726, -0.015901, -0.111475, -0.012635, -0.029585, -0.043885, -0.097188, -0.00683, 0.040953, -0.091716, -0.097066, 0.125877, 0.101466, -0.117703, -0.001483, 0.087936, -0.00806, 0.008189, 0.058145, -0.089602, 0.007832, 0.081087, -0.011696, 0.06454, 0.071723, 0.129059, 0.07864, 0.163052, 0.116826, -0.039481, -0.059053, -0.109745, 0.080791, 0.026902, -0.097094, -0.036005, -0.098446, 0.011695, -0.115028, 0.013473, -0.030081, 0.050539]}, "13": {"label": "wide-range · slow · question", "emb": [-0.049316, 0.079729, 0.134762, -0.017171, 0.016638, 0.03792, 0.077599, -0.047527, -0.058096, -0.125282, -0.065962, 0.054827, -0.042702, 0.080202, -0.128793, 0.047482, -0.015366, -0.128568, 0.174576, -0.064135, 0.028825, -0.057269, 0.013506, -0.084323, -0.018376, -0.006145, 0.135988, -0.051993, 0.045764, -0.068694, -0.096576, 0.060418, -0.033247, 0.118162, 0.214827, -0.079408, 0.013513, -0.143196, -0.043047, -0.070651, 0.008669, -0.048957, -0.056213, 0.003604, -0.062643, 0.060966, -0.091224, 0.131408, -0.051059, 0.116348, 0.082638, -0.183355, 0.106217, 0.073501, 0.069447, 0.046283, 0.073742, -0.094445, -0.096596, 0.057557, -0.018585, 0.11509, -0.091432, 0.025908]}, "14": {"label": "wide-range · normal", "emb": [0.001553, -0.012023, 0.022358, -0.114422, -0.004232, 0.085474, 0.091031, -0.031596, 0.057701, -0.040212, -0.030307, 0.059041, 0.029353, -0.018229, -0.066481, 0.040342, -0.084407, -0.071293, -0.011668, -0.019989, -0.097087, -0.081197, 0.043118, 0.05649, -0.095185, 0.082748, 0.037413, 0.053604, 0.048286, 0.028822, -0.014488, -0.034348, 0.006524, -0.035016, 0.036618, 0.001304, -0.009479, 0.160705, -0.002441, 0.008384, 0.018137, -0.126345, -0.07473, -0.048525, -0.066345, -0.034876, 0.039267, 0.043862, 0.02485, 0.089538, 0.044936, -0.034978, 0.019987, -0.038721, -0.009296, 0.00596, -0.083425, -0.002204, -0.127395, 0.020764, -0.041537, 0.060425, 0.03617, -0.042043]}, "15": {"label": "wide-range · normal · question", "emb": [0.031989, 0.12776, 0.097385, -0.115555, 0.019236, 0.075017, 0.006442, -0.068154, -0.075861, -0.122641, -0.011926, 0.1815, -0.024032, -0.059157, -0.206298, 0.076823, -0.15704, -0.082074, 0.253635, 0.075241, 0.020136, -0.032513, -0.070692, -0.050269, -0.057442, 0.141685, 0.191198, -0.020758, 0.109362, -0.140957, -0.093134, 0.079722, 0.104948, -0.135023, 0.117103, -0.141334, -0.067662, -0.041133, -0.069584, -0.123385, -0.084509, -0.171221, -0.072137, -0.091094, -0.012565, 0.058296, -0.145974, 0.03443, -0.038373, 0.012998, 0.006425, -0.059752, 0.043147, -0.073827, 0.020636, 0.113688, 0.142444, -0.03517, 0.023351, 0.126022, -0.010433, 0.027106, 0.032558, 0.033757]}, "16": {"label": "wide-range · fast", "emb": [0.115599, -0.161136, -0.017431, -0.194831, -0.190049, 0.115864, -0.027475, -0.059261, 0.004603, 0.053946, 0.058454, 0.123359, 0.167072, -0.035785, -0.042583, 0.098873, -0.055196, -0.016422, -0.073254, -0.057121, -0.117421, -0.035423, -0.049022, 0.022217, -0.045531, 0.14901, 0.03188, 0.039933, 0.114307, -0.010411, 0.067121, 0.07167, 0.077926, -0.039093, 0.018199, 0.13719, 0.112801, 0.193481, 0.011642, -0.087589, 0.05298, -0.081547, -0.037497, -0.120967, -0.089402, -0.108299, 0.098022, -0.054007, -0.052677, 0.008059, 0.001495, 0.017505, -0.021015, -0.057143, -0.132628, -0.041625, 0.020005, 0.046707, -0.255037, 0.038068, 0.124554, 0.037912, 0.14988, -0.066634]}, "17": {"label": "wide-range · fast · question", "emb": [0.144282, -0.070793, -0.002914, -0.1106, -0.186188, 0.1517, -0.046887, -0.091894, -0.099903, 0.008947, 0.039248, 0.194223, 0.059878, -0.129902, 0.013289, -0.02141, 0.007187, 0.010225, 0.133719, -0.010978, -0.103538, 0.001154, 0.001617, 0.02783, 0.174246, 0.161041, 0.069097, -0.014994, 0.140037, -0.069386, -0.149788, -0.034438, 0.068539, -0.06458, -0.006801, -0.000357, 0.067643, -0.222343, -0.03566, -0.223449, -0.043558, -0.130799, -0.064022, -0.154887, -0.037838, -0.038297, -0.081953, 0.020593, -0.169838, 0.140732, -0.054224, 0.050515, 0.021453, -0.180905, -0.079634, -0.081443, 0.068684, 0.10073, -0.081147, 0.084175, 0.024963, 0.050901, 0.029197, -0.016361]}}, "formula": "spk_emb = speakers['0'].emb (+ styles[k].emb for a style token; 7 = neutral)", "defaults": {"temperature": 0.8, "length_scale": 0.8, "style": 7}}
+{"default": "male_voxcpm_1", "targets": [], "presets": {"male_voxcpm_1": {"label": "male_voxcpm_1 (synthetic voice)", "weights": {"0": 1.0}}}, "speakers": {"0": {"name": "male_voxcpm_1", "emb": [-0.171126, 0.322866, -0.493196, -0.251865, -1.496564, -0.773552, 0.672265, -0.925804, -0.569764, -0.062601, 0.368567, -0.555231, -0.010228, 0.525713, 1.064089, 1.084695, -0.108082, -1.248569, 0.401326, -0.809678, 0.146417, 0.637237, 0.170811, -0.746055, 0.30452, -0.758564, -0.460776, -0.647524, -0.687801, 2.077753, -0.159227, -0.075347, 0.608661, 0.566903, -0.023465, -1.022551, -0.390619, -0.467591, 0.456134, 0.381035, -0.043261, -0.79233, -1.292068, -0.338539, 0.090066, -0.007928, 1.320292, -0.026984, -0.945175, -0.495162, -0.239323, 0.335876, -0.208299, 1.380403, -0.118773, -1.187929, 1.033132, 0.574047, 0.126154, 0.615193, -0.638149, -0.652656, 0.079094, 1.707609]}}, "styles": {"0": {"label": "flat-range · slow", "emb": [-0.145681, 0.028514, 0.031073, 0.213126, 0.141315, -0.091171, 0.020456, -0.050942, 0.116065, -0.037322, -0.050199, -0.128892, -0.080357, 0.045101, 0.012145, -0.013334, 0.032367, 0.00523, 0.016566, -0.002529, 0.094458, 0.054667, -7e-06, -0.011532, -0.006527, -0.130739, -0.04346, -0.025447, -0.056555, 0.02478, 0.035768, -0.007505, -0.064534, 0.114622, -0.005323, -0.011425, -0.081168, -0.026979, -0.054348, 0.063037, 0.021073, 0.113498, 0.125257, 0.107236, -0.032194, 0.021723, 0.069703, 0.023842, 0.102836, -0.123746, 0.012705, -0.015574, -0.023926, 0.049007, 0.077019, 0.142015, -0.022084, -0.036438, 0.126887, -0.074276, -0.087569, -0.009835, -0.031884, 0.1038]}, "1": {"label": "flat-range · slow · question", "emb": [-0.065793, 0.16722, -0.040421, 0.10364, 0.149176, -0.021905, -0.0432, -0.050239, -0.040979, -0.090244, -0.097514, -0.184097, -0.077265, 0.04736, -0.055627, -0.118585, 0.066956, -0.061624, 0.048341, 0.052993, 0.177459, 0.01611, -0.109275, -0.070324, 0.041396, 0.006175, 0.020192, -0.063335, 0.00341, -0.05934, -0.201707, 0.057273, 0.006712, 0.016891, 0.020629, -0.063877, -0.011998, -0.184993, -0.020477, 0.005055, -0.082237, 0.120278, -0.023095, 0.012626, 0.12684, 0.096845, -0.126149, 0.069004, -0.104394, 0.008861, 0.031055, -0.069648, 0.03642, 0.04356, 0.024144, 0.107632, 0.092163, 0.005802, 0.17045, -0.057356, -0.10544, 0.03624, -0.143395, 0.086159]}, "2": {"label": "flat-range · normal", "emb": [0.015198, 0.005387, -0.070077, 0.085694, 0.052682, -0.069297, -0.052193, -0.011528, 0.103583, 0.036779, -0.018627, -0.018627, -0.054331, -0.057561, 0.073174, 0.013283, 0.107597, 0.027577, 0.007558, 0.048043, -0.014568, 0.021438, -0.040362, -0.023572, 0.020275, -0.050076, -0.064918, 0.025708, -0.020152, -0.007477, 0.011547, 0.020923, -0.030765, -0.026945, -0.108283, 0.071716, -0.079583, -0.020836, 0.00641, 0.041376, -0.011752, 0.124817, 0.074644, 0.055393, 0.074216, 0.001747, 0.055778, -0.037595, 0.018188, -0.117185, -0.054989, 0.057521, 0.038923, 3.1e-05, -0.018487, 0.037477, 0.0652, 0.003342, 0.0622, 0.02653, 0.015227, -0.062551, 0.058737, -0.038891]}, "3": {"label": "flat-range · normal · question", "emb": [-0.033626, 0.123205, -0.104265, 0.116839, 0.158891, 0.03783, -0.124739, -0.13248, 0.047867, 0.023742, 0.102209, -0.103134, -0.015378, 0.050243, 0.184138, -0.05144, 0.137179, 0.063629, 0.10753, -0.026578, -0.051683, 0.090953, 0.096812, -0.112768, 0.075321, -0.093111, 0.098713, -0.018047, 0.111272, -0.088065, -0.118353, 0.12843, -0.041105, -0.091738, -0.013021, 0.096103, -0.080562, -0.139384, -0.049917, -0.024333, -0.139616, 0.064256, 0.010544, -0.032663, -0.035371, 0.009209, -0.197999, 0.041773, -0.06381, -0.09734, 0.00059, 0.013481, 0.011969, -0.06911, 0.026348, -0.020739, 0.100743, 0.028176, 0.141968, 0.027852, -0.026274, 0.098182, -0.024583, 0.015864]}, "4": {"label": "flat-range · fast", "emb": [0.079858, -0.032991, -0.141646, -0.0183, -0.05767, -0.010497, -0.130122, -0.063902, 0.006765, 0.108658, 0.042958, -0.02097, 0.050789, -0.10057, 0.101453, 0.113868, 0.042049, 0.099299, -0.084551, -0.002367, -0.086793, 0.08406, -0.099978, 0.061491, 0.097875, 0.108743, -0.111331, 0.084997, 0.144157, 0.00984, 0.065044, 0.041029, 0.064102, -0.095855, -0.145943, 0.078539, 0.045341, 0.036196, 0.068478, -0.028343, -0.042336, 0.131945, 0.046076, -0.006211, 0.078042, -0.122365, 0.007158, -0.175239, -0.052579, -0.182534, -0.110177, 0.126633, 0.037948, 0.037947, -0.06254, -0.004245, 0.158694, 0.066889, -0.018682, -0.003524, 0.125085, 0.024064, 0.130969, -0.106359]}, "5": {"label": "flat-range · fast · question", "emb": [0.09611, 0.034653, -0.220507, -0.148164, 0.024787, -0.047649, -0.211915, 0.173699, -0.072696, 0.068309, 0.094034, 0.072822, 0.019759, -0.108546, 0.158991, -0.032414, -0.033625, 0.035397, -0.038269, 0.020347, -0.026651, 0.06203, -0.057607, -0.056806, 0.105391, 0.188259, 0.159449, 0.088129, 0.173849, 0.078067, -0.155068, 0.117547, 0.10826, -0.139548, -0.019679, 0.050196, -0.005208, -0.135763, -0.048872, -0.156271, -0.068358, 0.079436, -0.067806, -0.098738, -0.000494, -0.008177, -0.207378, -0.158861, -0.088763, -0.091702, -0.054759, 0.171019, 0.123302, -0.038804, -0.038359, -0.009134, 0.188344, 0.213565, 0.065066, 0.131621, 0.000913, 0.027884, 0.028019, -0.047102]}, "6": {"label": "mid-range · slow", "emb": [-0.108414, -0.003766, 0.026759, 0.090972, 0.10488, 0.021303, 0.077811, 0.025143, 0.086748, -0.043222, -0.055501, -0.021596, -0.058954, 0.001334, 0.023127, -0.064402, -0.036275, -0.027464, -0.026106, -0.042851, 0.060475, -0.092045, 0.021875, 0.021948, -0.03574, -0.059152, -0.073984, 0.022593, -0.105647, 0.0545, 0.035674, -0.054579, -0.073088, 0.100104, 0.035317, -0.017922, -0.046948, -0.142279, -0.01218, 0.065619, 0.045547, -0.026938, 0.060306, 0.052985, 0.078487, 0.10227, 0.052383, 0.094831, 0.118608, 0.057531, 0.049244, -0.069276, 0.011932, -0.007232, 0.115472, 0.066428, -0.100617, -0.001337, 0.06223, -0.00322, -0.097148, -0.034642, -0.11133, 0.029218]}, "7": {"label": "mid-range · slow · question", "emb": [-0.029557, 0.133307, 0.098692, 0.122858, 0.09645, 0.008557, -0.014066, -0.135018, -0.079446, -0.12148, -0.135168, -0.129205, -0.064973, 0.125791, -0.074336, 0.01901, -0.015607, -0.046001, 0.083561, -0.01328, 0.075397, -0.012248, 0.024806, -0.042582, -0.084804, -0.042625, 0.019341, -0.129714, 0.029283, 0.029565, -0.080985, 0.070864, -0.059116, 0.07597, 0.106906, -0.044583, -0.017669, -0.20186, -0.141748, 0.010677, 0.008098, -0.023605, 0.056758, 0.098303, -0.034178, 0.045052, -0.183866, 0.117161, -0.002713, -0.024853, 0.015352, -0.082478, -0.014168, 0.110169, -0.000971, -0.01839, 0.054924, -0.169824, 0.123016, -0.045664, -0.173564, 0.084706, -0.13797, -0.031571]}, "8": {"label": "mid-range · normal", "emb": [0.047679, -0.033757, 0.012959, -0.020053, -0.017468, -0.00887, 0.010025, 0.062036, 0.099756, 0.013468, -0.017102, -0.037922, 0.03251, 0.001108, 0.008804, 0.02545, -9e-05, -0.001319, -0.010407, 0.004526, -0.03449, -0.04883, 0.001763, 0.010887, -0.014453, 0.03218, -0.055419, 0.059566, -0.003377, 0.022355, 0.007696, -0.022474, 0.013298, 0.010364, -0.023911, 0.041498, -0.037128, 0.003917, 0.013321, -0.010775, -0.001142, -0.059385, 0.010826, 0.003236, 0.048636, 0.028829, 0.040092, 0.028113, 0.009203, -0.027532, 0.003704, 0.009829, 0.002783, -0.042211, 0.032808, 0.036177, -0.001528, 0.017659, -0.00515, 0.016582, 0.046999, -0.007643, 0.016614, -0.050302]}, "9": {"label": "mid-range · normal · question", "emb": [0.100659, -0.035835, -0.122456, -0.055806, -0.030154, 0.026734, -0.065558, -0.071907, -0.035, -0.033529, -0.074552, -0.03608, -0.054054, -0.077936, -0.061722, -0.055407, 0.160748, 0.000847, -0.061431, 0.010338, 0.100741, 0.014006, -0.102683, -0.072087, 0.03347, 0.006653, 0.183114, -0.049406, -0.037271, 0.041016, -0.186704, -0.042957, 0.027818, 0.038135, 0.052061, 0.089554, 0.044856, -0.021578, -0.071574, 0.037154, -0.032053, -0.05052, -0.023911, -0.050673, -0.013418, 0.040641, -0.213389, 0.059986, -0.13572, 0.073084, -0.129966, -0.05335, 0.188839, 0.034079, -0.022861, -0.074348, 0.077725, -0.061973, -0.01645, 0.041444, -0.007721, 0.011829, -0.024506, 0.00019]}, "10": {"label": "mid-range · fast", "emb": [0.148559, -0.117011, -0.108052, -0.079113, -0.151191, -0.009634, -0.05918, 0.006944, 0.0017, 0.115003, 0.033147, 0.08128, 0.118907, -0.075013, -0.014178, 0.08572, -0.014366, 0.092867, -0.012872, -0.028587, -0.077081, 0.010422, -0.063002, 0.000577, -0.009214, 0.135065, -0.056282, 0.013967, 0.067966, -0.058367, 0.023978, 0.070172, 0.075121, -0.092142, -0.06845, 0.120971, 0.014942, 0.059432, 0.032089, -0.000947, -0.023865, -0.014222, 0.013042, -0.099835, -0.027346, -0.049702, 0.015537, -0.066083, -0.075356, -0.094995, -0.060072, 0.054041, 0.036734, -0.039968, -0.078773, -0.020267, -0.01236, 0.055531, -0.081482, 0.057737, 0.08034, -0.001431, 0.14718, -0.100518]}, "11": {"label": "mid-range · fast · question", "emb": [0.073619, 0.043859, -0.05577, -0.034364, -0.257909, 0.070455, -0.16292, -0.010065, -0.044395, 0.102808, -0.050129, 0.0522, 0.193897, -0.029656, 0.007619, 0.052111, 0.04524, 0.049192, 0.113122, 0.129318, -0.029497, 0.10029, -0.2246, 0.008363, 0.124819, 0.078415, 0.031787, -0.101723, 0.063401, -0.14017, -0.119636, 0.044516, 0.090903, -0.005523, -0.089667, 0.139016, 0.156393, 0.024767, -0.054849, -0.176828, -0.139236, -0.009136, -0.24576, -0.182452, -0.160701, -0.003506, -0.06909, -0.035963, -0.067486, -0.004653, 0.015554, 0.119868, 0.013078, -0.071847, -0.064107, -0.047181, 0.090308, 0.078807, 0.023879, 0.118498, 0.09614, 0.184368, -0.05321, -0.100499]}, "12": {"label": "wide-range · slow", "emb": [-0.11099, 0.018829, 0.151204, 0.0139, 0.021509, 0.077328, 0.178788, -0.00928, 0.037616, -0.079435, -0.026652, 0.004519, -0.03194, -0.02984, -0.083479, 0.028554, -0.129474, -0.073987, 0.002037, -0.035017, 0.00193, -0.117057, 0.072726, -0.015901, -0.111475, -0.012635, -0.029585, -0.043885, -0.097188, -0.00683, 0.040953, -0.091716, -0.097066, 0.125877, 0.101466, -0.117703, -0.001483, 0.087936, -0.00806, 0.008189, 0.058145, -0.089602, 0.007832, 0.081087, -0.011696, 0.06454, 0.071723, 0.129059, 0.07864, 0.163052, 0.116826, -0.039481, -0.059053, -0.109745, 0.080791, 0.026902, -0.097094, -0.036005, -0.098446, 0.011695, -0.115028, 0.013473, -0.030081, 0.050539]}, "13": {"label": "wide-range · slow · question", "emb": [-0.049316, 0.079729, 0.134762, -0.017171, 0.016638, 0.03792, 0.077599, -0.047527, -0.058096, -0.125282, -0.065962, 0.054827, -0.042702, 0.080202, -0.128793, 0.047482, -0.015366, -0.128568, 0.174576, -0.064135, 0.028825, -0.057269, 0.013506, -0.084323, -0.018376, -0.006145, 0.135988, -0.051993, 0.045764, -0.068694, -0.096576, 0.060418, -0.033247, 0.118162, 0.214827, -0.079408, 0.013513, -0.143196, -0.043047, -0.070651, 0.008669, -0.048957, -0.056213, 0.003604, -0.062643, 0.060966, -0.091224, 0.131408, -0.051059, 0.116348, 0.082638, -0.183355, 0.106217, 0.073501, 0.069447, 0.046283, 0.073742, -0.094445, -0.096596, 0.057557, -0.018585, 0.11509, -0.091432, 0.025908]}, "14": {"label": "wide-range · normal", "emb": [0.001553, -0.012023, 0.022358, -0.114422, -0.004232, 0.085474, 0.091031, -0.031596, 0.057701, -0.040212, -0.030307, 0.059041, 0.029353, -0.018229, -0.066481, 0.040342, -0.084407, -0.071293, -0.011668, -0.019989, -0.097087, -0.081197, 0.043118, 0.05649, -0.095185, 0.082748, 0.037413, 0.053604, 0.048286, 0.028822, -0.014488, -0.034348, 0.006524, -0.035016, 0.036618, 0.001304, -0.009479, 0.160705, -0.002441, 0.008384, 0.018137, -0.126345, -0.07473, -0.048525, -0.066345, -0.034876, 0.039267, 0.043862, 0.02485, 0.089538, 0.044936, -0.034978, 0.019987, -0.038721, -0.009296, 0.00596, -0.083425, -0.002204, -0.127395, 0.020764, -0.041537, 0.060425, 0.03617, -0.042043]}, "15": {"label": "wide-range · normal · question", "emb": [0.031989, 0.12776, 0.097385, -0.115555, 0.019236, 0.075017, 0.006442, -0.068154, -0.075861, -0.122641, -0.011926, 0.1815, -0.024032, -0.059157, -0.206298, 0.076823, -0.15704, -0.082074, 0.253635, 0.075241, 0.020136, -0.032513, -0.070692, -0.050269, -0.057442, 0.141685, 0.191198, -0.020758, 0.109362, -0.140957, -0.093134, 0.079722, 0.104948, -0.135023, 0.117103, -0.141334, -0.067662, -0.041133, -0.069584, -0.123385, -0.084509, -0.171221, -0.072137, -0.091094, -0.012565, 0.058296, -0.145974, 0.03443, -0.038373, 0.012998, 0.006425, -0.059752, 0.043147, -0.073827, 0.020636, 0.113688, 0.142444, -0.03517, 0.023351, 0.126022, -0.010433, 0.027106, 0.032558, 0.033757]}, "16": {"label": "wide-range · fast", "emb": [0.115599, -0.161136, -0.017431, -0.194831, -0.190049, 0.115864, -0.027475, -0.059261, 0.004603, 0.053946, 0.058454, 0.123359, 0.167072, -0.035785, -0.042583, 0.098873, -0.055196, -0.016422, -0.073254, -0.057121, -0.117421, -0.035423, -0.049022, 0.022217, -0.045531, 0.14901, 0.03188, 0.039933, 0.114307, -0.010411, 0.067121, 0.07167, 0.077926, -0.039093, 0.018199, 0.13719, 0.112801, 0.193481, 0.011642, -0.087589, 0.05298, -0.081547, -0.037497, -0.120967, -0.089402, -0.108299, 0.098022, -0.054007, -0.052677, 0.008059, 0.001495, 0.017505, -0.021015, -0.057143, -0.132628, -0.041625, 0.020005, 0.046707, -0.255037, 0.038068, 0.124554, 0.037912, 0.14988, -0.066634]}, "17": {"label": "wide-range · fast · question", "emb": [0.144282, -0.070793, -0.002914, -0.1106, -0.186188, 0.1517, -0.046887, -0.091894, -0.099903, 0.008947, 0.039248, 0.194223, 0.059878, -0.129902, 0.013289, -0.02141, 0.007187, 0.010225, 0.133719, -0.010978, -0.103538, 0.001154, 0.001617, 0.02783, 0.174246, 0.161041, 0.069097, -0.014994, 0.140037, -0.069386, -0.149788, -0.034438, 0.068539, -0.06458, -0.006801, -0.000357, 0.067643, -0.222343, -0.03566, -0.223449, -0.043558, -0.130799, -0.064022, -0.154887, -0.037838, -0.038297, -0.081953, 0.020593, -0.169838, 0.140732, -0.054224, 0.050515, 0.021453, -0.180905, -0.079634, -0.081443, 0.068684, 0.10073, -0.081147, 0.084175, 0.024963, 0.050901, 0.029197, -0.016361]}}, "formula": "spk_emb = speakers['0'].emb (+ styles[k].emb for a style token; 7 = neutral)", "defaults": {"temperature": 1.0, "length_scale": 0.8, "style": 7}}

samples/README.md

@@ -1,4 +1,4 @@
-Held-out test sentences (not in the training set). v2 at the default settings (tempo 0.8, temperature 0.8, 4 ODE steps); teacher = VoxCPM2 with the frozen male_voxcpm_1 voice.
+Held-out test sentences (not in the training set). v2 at tempo 0.8, temperature 0.8, 4 ODE steps (the Space default temperature is now 1.0); teacher = VoxCPM2 with the frozen male_voxcpm_1 voice.
 
 | # | text | v2 | VoxCPM2 teacher |
 |---|---|---|---|