Alania-2 production row: best of 11 voices (12 Oct 2026, alania-v2.1)

Re-ran Alania-2 through the production API for all 11 built-in voices with the same harness and scoring; the row shows the best voice per column (WER/CER: Öncel, clean/UTMOS: Lina), named, with a footnote. Adds those two voices' per-clip scores.

README.md

@@ -23,7 +23,7 @@ an 8 kHz band, faster-whisper large-v3 at beam 5, and EMA's `score_norm`. Speed
 
 **Charts, leaderboard and audio samples:** https://speech.patientdesk.ai/en/benchmarks (Turkish: https://speech.patientdesk.ai/benchmarks)
 
-Run on 7 October 2026 by PatientDesk AI. Alania-2 and Alania-1 are our commercial models; Antalia-1 is our open-weights release.
+Run on 7 October 2026 by PatientDesk AI; the Alania-2 production row on 12 October. Alania-2 and Alania-1 are our commercial models; Antalia-1 is our open-weights release.
 
 ## Results
 
@@ -31,13 +31,13 @@ Run on 7 October 2026 by PatientDesk AI. Alania-2 and Alania-1 are our commercia
 |---|---|---|---|---|---|---|---|---|
 | 1 | EMA Lightning (plain PyTorch) | 8.6M | 1.04 ± 0.17 | 0.22 | 93.7% | 3.30 | 0.017 | ours |
 | 2 | EMA Lightning (compiled) | 8.6M | 1.04 ± 0.17 | 0.22 | 93.7% | 3.30 | 0.0030 | ours |
-| 3 | **Antalia-2 Mini** (open weights, default settings) | 7.62M | 1.05 ± 0.07 | 0.20 | 93.1% | 3.25 | 0.043 | ours |
-| 4 | Trendyol-TTS (cfg 2.0, 16 steps) | 2.38B | 1.16 ± 0.16 | 0.23 | 93.1% | 3.82 | 0.962 | ours |
-| 5 | Gemini 3.8 Flash TTS (voice Kore) | – | 1.23 ± 0.10 | 0.29 | 92.9% | 3.53 | – | ours |
-| 6 | Gemini 3.8 Flash-Lite TTS (voice Kore) | – | 1.34 ± 0.19 | 0.27 | 91.9% | 3.63 | – | ours |
-| 7 | Anka TTS (built-in male voice) | 336M | 1.42 ± 0.06 | 0.30 | 91.6% | 3.52 | 0.322 | ours |
-| 8 | ElevenLabs v4 (voice George) | – | 1.43 ± 0.12 | 0.27 | – | – | – | quoted: EMA Lightning card |
-| 9 | **Alania-2** (production API, voice F1) | 488M | 1.44 ± 0.15 | 0.28 | 91.7% | 3.67 | – | ours |
+| 3 | **Antalia-2 Mini 1.0.1** (open weights, default settings) | 7.62M | 1.09 ± 0.12 | 0.22 | 92.9% | 3.18 | 0.029 | ours |
+| 4 | **Alania-2** (production API, best of 11 voices¹) | 488M | 1.15 ± 0.22 (Öncel) | 0.22 (Öncel) | 93.1% (Lina) | 3.66 (Lina) | – | ours |
+| 5 | Trendyol-TTS (cfg 2.0, 16 steps) | 2.38B | 1.16 ± 0.16 | 0.23 | 93.1% | 3.82 | 0.962 | ours |
+| 6 | Gemini 3.8 Flash TTS (voice Kore) | – | 1.23 ± 0.10 | 0.29 | 92.9% | 3.53 | – | ours |
+| 7 | Gemini 3.8 Flash-Lite TTS (voice Kore) | – | 1.34 ± 0.19 | 0.27 | 91.9% | 3.63 | – | ours |
+| 8 | Anka TTS (built-in male voice) | 336M | 1.42 ± 0.06 | 0.30 | 91.6% | 3.52 | 0.322 | ours |
+| 9 | ElevenLabs v4 (voice George) | – | 1.43 ± 0.12 | 0.27 | – | – | – | quoted: EMA Lightning card |
 | 10 | **Alania-2** (model only, no serving checks) | 488M | 1.51 ± 0.19 | 0.30 | 91.8% | 3.66 | 0.151 | ours |
 | 11 | ElevenLabs v4 Turbo (voice George) | – | 1.54 ± 0.22 | 0.29 | – | – | – | quoted: EMA Lightning card |
 | 12 | Chatterbox Multilingual (Anka male reference) | 500M | 1.91 ± 0.35 | 0.43 | 88.8% | 3.51 | 2.892 | ours |
@@ -50,10 +50,14 @@ Run on 7 October 2026 by PatientDesk AI. Alania-2 and Alania-1 are our commercia
 | 19 | FreyaTTS-small (default voice) | 183M | 12.02 ± 0.00 | 4.95 | 47.1% | 2.81 | 0.286 | ours |
 | 20 | F5-TTS base (no Turkish training) | 336M | 77.84 ± 0.13 | 32.58 | – | – | – | quoted: Anka TTS card |
 
+¹ Alania-2 was re-run on 12 October 2026 (release alania-v2.1, production API) for all 11 built-in voices with the same harness and scoring; each column shows the best voice: Öncel for WER and CER, Lina for clean rate and UTMOS. The 11 are within noise of each other (WER 1.15–1.62%); the default voice F1 alone scored 1.38 ± 0.18 (1.44 ± 0.15 on 7 October). The other rows are from 7 October.
+
 WER/CER: corpus level per take after `score_norm`, mean ± sample standard deviation over three takes. Clean: share of sentences
 with no word error. UTMOS: UTMOS22 strong on full-band audio (English-trained; a rough guide only). RTF: generation time / audio
 length on one NVIDIA L4, middle of three passes after a warm-up; API rows have none. Alania-1's RTF (2.49 on the L4, model only)
-was measured on the first 100 sentences. Chatterbox Multilingual slowed down from pass to pass (RTF 1.69, 2.89, 4.21) with no change in output length; as for every system, the middle pass is reported. Rows marked *quoted* were not run by us.
+was measured on the first 100 sentences. Antalia-2 Mini was re-run on 10 October 2026 at 1.0.1, which changed only its
+output EQ (1.0.0: 1.05 ± 0.07 / 0.20 / 93.1% / 3.25 / 0.043); its CER is 0.220 against EMA Lightning's 0.225, which both
+round to 0.22. Chatterbox Multilingual slowed down from pass to pass (RTF 1.69, 2.89, 4.21) with no change in output length; as for every system, the middle pass is reported. Rows marked *quoted* were not run by us.
 
 ## What is in this repo
 
@@ -79,7 +83,7 @@ the sentences are fetched from the Hub.
 | System | Install | Run |
 |---|---|---|
 | EMA Lightning 1.0.1 | `pip install ema-lightning soundfile` | `run_ema.py plain` / `run_ema.py compiled` |
-| Antalia-2 Mini 1.0.0 | `pip install <wheel from huggingface.co/cloud0day3/antalia-mini>` | `run_antalia_mini.py` |
+| Antalia-2 Mini 1.0.1 | `pip install <wheel from huggingface.co/cloud0day3/antalia-mini>` | `run_antalia_mini.py` |
 | Trendyol-TTS | `pip install voxcpm soundfile` | `run_trendyol.py` |
 | Anka TTS v0.1 | `pip install "anka-tts[tts]" "pyarrow<17" soundfile` | `run_anka.py` |
 | Chatterbox Multilingual | `pip install chatterbox-tts "setuptools<81" soundfile` | `REF_WAV=ref/male.wav run_chatterbox.py` |

results/results.json

@@ -837,20 +837,20 @@
       {
         "take": "take0",
         "n": 495,
-        "wer": 1.1233086545825888,
-        "cer": 0.21131459924371618,
-        "clean": 92.72727272727273,
-        "utmos": 3.256870967691595,
+        "wer": 1.225427623181006,
+        "cer": 0.23726551494031287,
+        "clean": 92.32323232323232,
+        "utmos": 3.190684290606566,
         "audio_s": 1705.7386666666646,
         "silent": 0
       },
       {
         "take": "take1",
         "n": 495,
-        "wer": 1.0211896859841716,
-        "cer": 0.18907095721806183,
-        "clean": 93.33333333333333,
-        "utmos": 3.241021330669673,
+        "wer": 1.0467194281337757,
+        "cer": 0.21131459924371618,
+        "clean": 93.13131313131314,
+        "utmos": 3.174159880840417,
         "audio_s": 1705.7386666666646,
         "silent": 0
       },
@@ -858,27 +858,125 @@
         "take": "take2",
         "n": 495,
         "wer": 0.9956599438345672,
-        "cer": 0.20019277823088902,
+        "cer": 0.21131459924371618,
         "clean": 93.33333333333333,
-        "utmos": 3.2445005151960586,
+        "utmos": 3.18210461211927,
         "audio_s": 1705.7386666666646,
         "silent": 0
       }
     ],
     "n_takes": 3,
-    "wer": 1.046719428133776,
-    "wer_sd": 0.06754534876345653,
+    "wer": 1.0892689983831163,
+    "wer_sd": 0.12064885438479989,
     "wer_min": 0.9956599438345672,
-    "wer_max": 1.1233086545825888,
-    "cer": 0.20019277823088902,
-    "cer_sd": 0.011121821012827174,
-    "clean": 93.13131313131314,
-    "utmos": 3.2474642711857755,
-    "always_right": 449,
+    "wer_max": 1.225427623181006,
+    "cer": 0.21996490447591507,
+    "cer_sd": 0.014982768163147383,
+    "clean": 92.92929292929293,
+    "utmos": 3.182316261188751,
+    "always_right": 445,
     "always_wrong": 24,
-    "rtf": 0.04343431939886861,
+    "rtf": 0.028583162304267024,
     "rtf_n": 495,
     "gpu": "NVIDIA L4",
     "sample_rate": 48000
+  },
+  "alania-2-oncel": {
+    "takes": [
+      {
+        "take": "take0",
+        "n": 495,
+        "wer": 0.8935409752361502,
+        "cer": 0.16682731519240748,
+        "clean": 94.34343434343434,
+        "utmos": 3.5155335190320254,
+        "audio_s": 1636.32,
+        "silent": 0
+      },
+      {
+        "take": "take1",
+        "n": 495,
+        "wer": 1.2764871074802144,
+        "cer": 0.24468006228219766,
+        "clean": 92.32323232323232,
+        "utmos": 3.5358553949028555,
+        "audio_s": 1633.66,
+        "silent": 0
+      },
+      {
+        "take": "take2",
+        "n": 495,
+        "wer": 1.2764871074802144,
+        "cer": 0.24097278861125526,
+        "clean": 92.32323232323232,
+        "utmos": 3.5125094033250903,
+        "audio_s": 1636.48,
+        "silent": 0
+      }
+    ],
+    "n_takes": 3,
+    "wer": 1.148838396732193,
+    "wer_sd": 0.22109405253623654,
+    "wer_min": 0.8935409752361502,
+    "wer_max": 1.2764871074802144,
+    "cer": 0.2174933886952868,
+    "cer_sd": 0.04391724288553635,
+    "clean": 92.996632996633,
+    "utmos": 3.521299439086657,
+    "always_right": 438,
+    "always_wrong": 20,
+    "rtf": null,
+    "rtf_n": null,
+    "gpu": null,
+    "sample_rate": 48000
+  },
+  "alania-2-lina": {
+    "takes": [
+      {
+        "take": "take0",
+        "n": 495,
+        "wer": 1.2764871074802144,
+        "cer": 0.24838733595314005,
+        "clean": 92.92929292929293,
+        "utmos": 3.6655750332456645,
+        "audio_s": 1761.1200000000001,
+        "silent": 0
+      },
+      {
+        "take": "take1",
+        "n": 495,
+        "wer": 1.1488383967321931,
+        "cer": 0.2520946096240824,
+        "clean": 93.53535353535354,
+        "utmos": 3.6657491804373383,
+        "audio_s": 1760.48,
+        "silent": 0
+      },
+      {
+        "take": "take2",
+        "n": 495,
+        "wer": 1.327546591779423,
+        "cer": 0.2743382516497368,
+        "clean": 92.92929292929293,
+        "utmos": 3.6479770010167902,
+        "audio_s": 1763.3600000000001,
+        "silent": 0
+      }
+    ],
+    "n_takes": 3,
+    "wer": 1.25095736533061,
+    "wer_sd": 0.0920487943697724,
+    "wer_min": 1.1488383967321931,
+    "wer_max": 1.327546591779423,
+    "cer": 0.2582733990756531,
+    "cer_sd": 0.014035511593386917,
+    "clean": 93.13131313131314,
+    "utmos": 3.6597670715665975,
+    "always_right": 443,
+    "always_wrong": 20,
+    "rtf": null,
+    "rtf_n": null,
+    "gpu": null,
+    "sample_rate": 48000
   }
 }