Add files using upload-large-folder tool

LICENSE

@@ -0,0 +1,12 @@
+Creative Commons Attribution 4.0 International
+
+Copyright (c) 2026 Muhammed Kumcu and Yagmur Tuncer
+
+The benchmark metrics and original documentation in this repository are licensed
+under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
+
+This license does not apply to third-party source datasets, audio, transcripts,
+model weights, or model outputs. Those materials remain subject to their original
+licenses and terms.
+
+License text: https://creativecommons.org/licenses/by/4.0/legalcode

README.md

@@ -0,0 +1,103 @@
+---
+license: cc-by-4.0
+language:
+- tr
+task_categories:
+- automatic-speech-recognition
+tags:
+- turkish
+- speech-recognition
+- benchmark
+- wer
+- cer
+- real-time-factor
+pretty_name: TurkMedSTT Turkish ASR Benchmark
+size_categories:
+- 10K<n<100K
+---
+
+# TurkMedSTT Turkish ASR Benchmark
+
+This repository publishes general-domain Turkish ASR benchmark metrics produced by
+Muhammed Kumcu and Yagmur Tuncer.
+
+The release compares 20 ASR models on 1,060 clips (about 105 minutes) drawn from
+three established Turkish speech sources. It contains metrics only. Source audio,
+reference transcripts, generated hypotheses, local paths, and medical evaluation
+results are not distributed.
+
+## Dataset composition
+
+| Source | Clips |
+|---|---:|
+| Common Voice Turkish | 447 |
+| ISSAI Turkish | 453 |
+| OpenSLR 106 Turkish | 160 |
+| **Total** | **1,060** |
+
+## Results
+
+WER and CER are macro means over clips. RTF is the mean real-time factor measured
+in the original Colab A100 runs.
+
+| Rank | Model | Clips | WER | CER | RTF |
+|---:|---|---:|---:|---:|---:|
+| 1 | `openai/whisper-large-v3` | 1,060 | 0.1345 | 0.0588 | 0.1349 |
+| 2 | `vincespeed/faster-whisper-large-v3-turbo-turkish` | 1,060 | 0.1825 | 0.0774 | 0.0927 |
+| 3 | `Huseyin/whisper-large-v3-turkish-finetuned` | 1,060 | 0.1841 | 0.0971 | 0.1587 |
+| 4 | `openai/whisper-large-v2` | 1,060 | 0.1901 | 0.0999 | 0.1333 |
+| 5 | `openai/whisper-large-v3-turbo` | 1,060 | 0.2014 | 0.0897 | 0.0338 |
+
+The complete 20-model ranking is in `summary/leaderboard.csv`. Per-source WER,
+CER, and RTF are in `summary/source_breakdown.csv`.
+
+## AcoSemantic-TR
+
+The release also includes ASCS_text, an experimental text-level semantic-affective
+consistency score. It is reported as an additional analysis dimension and must not
+be interpreted as a replacement for WER or CER.
+
+`facebook/seamless-m4t-v2-large` has the highest mean ASCS_text (0.9609), while
+`openai/whisper-large-v3` has the best WER and a mean ASCS_text of 0.9571.
+
+## Files
+
+- `summary/leaderboard.csv`: overall WER, CER, RTF, runtime, and source counts.
+- `summary/source_breakdown.csv`: metrics by model and source.
+- `summary/acosemantic_summary.csv`: model-level ASCS_text summary.
+- `data/per_utterance_metrics.csv`: de-identified clip-level WER, CER, RTF, and timing.
+- `data/acosemantic_per_utterance.csv`: de-identified clip-level ASCS_text values.
+
+Audio identifiers are retained only to join metrics across tables. No audio or
+transcript content is included.
+
+## Scope and limitations
+
+- These results cover general-domain Turkish speech only.
+- Results are hardware- and implementation-dependent, especially RTF.
+- Some original model attempts required dependency or loader repairs before the
+  final valid runs.
+- Source datasets remain governed by their original licenses and terms.
+- This metrics release does not grant rights to redistribute the source audio or
+  transcripts.
+
+## Contributors
+
+- Muhammed Kumcu - equal contributor
+- Yagmur Tuncer - equal contributor
+
+## Acknowledgements
+
+We thank Zeynep Zehra Kumcu and Yusuf Uysal for their support.
+
+## Citation
+
+```bibtex
+@misc{kumcu_tuncer_2026_turkish_asr_benchmark,
+  title        = {TurkMedSTT Turkish ASR Benchmark},
+  author       = {Kumcu, Muhammed and Tuncer, Yagmur},
+  year         = {2026},
+  publisher    = {Hugging Face},
+  organization = {TurkMedSTT}
+}
+```

summary/acosemantic_summary.csv

@@ -0,0 +1,21 @@
+model,clips,mean_wer,mean_ascs_text
+openai/whisper-large-v3,1060,0.134504,0.957067
+vincespeed/faster-whisper-large-v3-turbo-turkish,1060,0.182487,0.953372
+Huseyin/whisper-large-v3-turkish-finetuned,1060,0.184089,0.951012
+openai/whisper-large-v2,1060,0.190054,0.949753
+openai/whisper-large-v3-turbo,1060,0.201426,0.948699
+facebook/seamless-m4t-v2-large,1060,0.207889,0.960871
+openai/whisper-medium,1060,0.214342,0.947561
+mpoyraz/wav2vec2-xls-r-300m-cv8-turkish,1060,0.290263,0.941828
+openai/whisper-small,1060,0.300016,0.944819
+facebook/mms-1b-fl102,1060,0.461160,0.929665
+ceyda/wav2vec2-large-xlsr-53-turkish,1060,0.483158,0.932044
+m3hrdadfi/wav2vec2-large-xlsr-turkish,1060,0.484739,0.931591
+cahya/wav2vec2-large-xlsr-turkish,1060,0.521818,0.932345
+mbsouksu/wav2vec2-large-xlsr-turkish-large,1060,0.522395,0.929317
+openai/whisper-base,1060,0.523847,0.932433
+selimc/whisper-large-v3-turbo-turkish,1060,0.538217,0.951023
+Baybars/wav2vec2-xls-r-300m-cv8-turkish,1060,0.554992,0.929746
+erenfazlioglu/whisper-small-turkish-tr-best,1060,0.619535,0.951523
+openai/whisper-tiny,1060,0.654741,0.917986
+gorkemgoknar/wav2vec2-large-xlsr-53-turkish,1060,0.675864,0.924217

summary/leaderboard.csv

@@ -0,0 +1,21 @@
+rank,model,backend,clips,mean_wer,mean_cer,mean_rtf,runtime_seconds,audio_minutes,commonvoice_tr_clips,issai_clips,openslr_tr_clips
+1,openai/whisper-large-v3,transformers,1060,0.134504,0.058806,0.134905,855.836569,104.979336,447,453,160
+2,vincespeed/faster-whisper-large-v3-turbo-turkish,faster-whisper,1060,0.182487,0.077418,0.092666,445.826188,104.979336,447,453,160
+3,Huseyin/whisper-large-v3-turkish-finetuned,transformers,1060,0.184089,0.097132,0.158708,970.019427,104.979336,447,453,160
+4,openai/whisper-large-v2,transformers,1060,0.190054,0.099924,0.133312,848.966390,104.979336,447,453,160
+5,openai/whisper-large-v3-turbo,transformers,1060,0.201426,0.089659,0.033790,201.361660,104.979336,447,453,160
+6,facebook/seamless-m4t-v2-large,transformers_pipeline,1060,0.207889,0.096070,0.284810,1503.776660,104.979336,447,453,160
+7,openai/whisper-medium,openai-whisper,1060,0.214342,0.106904,0.215078,1287.088459,104.979336,447,453,160
+8,mpoyraz/wav2vec2-xls-r-300m-cv8-turkish,transformers_ctc,1060,0.290263,0.066539,0.125610,584.158747,104.979336,447,453,160
+9,openai/whisper-small,openai-whisper,1060,0.300016,0.132711,0.117789,668.649329,104.979336,447,453,160
+10,facebook/mms-1b-fl102,transformers_ctc,1060,0.461160,0.145392,0.374402,2105.209502,104.979336,447,453,160
+11,ceyda/wav2vec2-large-xlsr-53-turkish,transformers_ctc,1060,0.483158,0.111762,0.125375,755.677303,104.979336,447,453,160
+12,m3hrdadfi/wav2vec2-large-xlsr-turkish,transformers_ctc,1060,0.484739,0.110195,0.123642,740.922085,104.979336,447,453,160
+13,cahya/wav2vec2-large-xlsr-turkish,transformers_ctc,1060,0.521818,0.140439,0.155390,1018.364534,104.979336,447,453,160
+14,mbsouksu/wav2vec2-large-xlsr-turkish-large,transformers_ctc,1060,0.522395,0.126566,0.207319,1256.698968,104.979336,447,453,160
+15,openai/whisper-base,openai-whisper,1060,0.523847,0.229115,0.172582,782.138997,104.979336,447,453,160
+16,selimc/whisper-large-v3-turbo-turkish,transformers,1060,0.538217,0.348104,0.037401,209.364705,104.979336,447,453,160
+17,Baybars/wav2vec2-xls-r-300m-cv8-turkish,transformers_ctc,1060,0.554992,0.139311,0.212968,1264.883364,104.979336,447,453,160
+18,erenfazlioglu/whisper-small-turkish-tr-best,transformers,1060,0.619535,0.349951,0.084317,547.053765,104.979336,447,453,160
+19,openai/whisper-tiny,openai-whisper,1060,0.654741,0.315355,0.108289,527.961855,104.979336,447,453,160
+20,gorkemgoknar/wav2vec2-large-xlsr-53-turkish,transformers_ctc,1060,0.675864,0.207424,0.122976,741.012614,104.979336,447,453,160

summary/source_breakdown.csv

@@ -0,0 +1,61 @@
+model,backend,source,clips,mean_wer,mean_cer,mean_rtf
+Baybars/wav2vec2-xls-r-300m-cv8-turkish,transformers_ctc,commonvoice_tr,447,0.476378,0.115185,0.233771
+Baybars/wav2vec2-xls-r-300m-cv8-turkish,transformers_ctc,issai,453,0.597377,0.151288,0.203830
+Baybars/wav2vec2-xls-r-300m-cv8-turkish,transformers_ctc,openslr_tr,160,0.654613,0.172800,0.180725
+cahya/wav2vec2-large-xlsr-turkish,transformers_ctc,commonvoice_tr,447,0.375586,0.076191,0.142060
+cahya/wav2vec2-large-xlsr-turkish,transformers_ctc,issai,453,0.616092,0.176155,0.156448
+cahya/wav2vec2-large-xlsr-turkish,transformers_ctc,openslr_tr,160,0.663440,0.218809,0.189633
+ceyda/wav2vec2-large-xlsr-53-turkish,transformers_ctc,commonvoice_tr,447,0.392623,0.088238,0.132695
+ceyda/wav2vec2-large-xlsr-53-turkish,transformers_ctc,issai,453,0.545419,0.128157,0.123613
+ceyda/wav2vec2-large-xlsr-53-turkish,transformers_ctc,openslr_tr,160,0.559816,0.131065,0.109912
+erenfazlioglu/whisper-small-turkish-tr-best,transformers,commonvoice_tr,447,0.229669,0.112424,0.063088
+erenfazlioglu/whisper-small-turkish-tr-best,transformers,issai,453,1.030517,0.579215,0.100555
+erenfazlioglu/whisper-small-turkish-tr-best,transformers,openslr_tr,160,0.545129,0.364435,0.097649
+facebook/mms-1b-fl102,transformers_ctc,commonvoice_tr,447,0.414497,0.133553,0.495589
+facebook/mms-1b-fl102,transformers_ctc,issai,453,0.521073,0.164827,0.301524
+facebook/mms-1b-fl102,transformers_ctc,openslr_tr,160,0.421895,0.123445,0.242174
+facebook/seamless-m4t-v2-large,transformers_pipeline,commonvoice_tr,447,0.231068,0.111527,0.290559
+facebook/seamless-m4t-v2-large,transformers_pipeline,issai,453,0.224104,0.102882,0.320352
+facebook/seamless-m4t-v2-large,transformers_pipeline,openslr_tr,160,0.097219,0.033601,0.168121
+gorkemgoknar/wav2vec2-large-xlsr-53-turkish,transformers_ctc,commonvoice_tr,447,0.594412,0.150785,0.129987
+gorkemgoknar/wav2vec2-large-xlsr-53-turkish,transformers_ctc,issai,453,0.732018,0.261676,0.121316
+gorkemgoknar/wav2vec2-large-xlsr-53-turkish,transformers_ctc,openslr_tr,160,0.744433,0.212056,0.108093
+Huseyin/whisper-large-v3-turkish-finetuned,transformers,commonvoice_tr,447,0.168450,0.082703,0.134161
+Huseyin/whisper-large-v3-turkish-finetuned,transformers,issai,453,0.161879,0.075731,0.184054
+Huseyin/whisper-large-v3-turkish-finetuned,transformers,openslr_tr,160,0.290659,0.198034,0.155525
+m3hrdadfi/wav2vec2-large-xlsr-turkish,transformers_ctc,commonvoice_tr,447,0.387081,0.080961,0.132013
+m3hrdadfi/wav2vec2-large-xlsr-turkish,transformers_ctc,issai,453,0.553476,0.129969,0.121124
+m3hrdadfi/wav2vec2-large-xlsr-turkish,transformers_ctc,openslr_tr,160,0.562959,0.135884,0.107382
+mbsouksu/wav2vec2-large-xlsr-turkish-large,transformers_ctc,commonvoice_tr,447,0.430778,0.095546,0.209817
+mbsouksu/wav2vec2-large-xlsr-turkish-large,transformers_ctc,issai,453,0.591251,0.149037,0.210391
+mbsouksu/wav2vec2-large-xlsr-turkish-large,transformers_ctc,openslr_tr,160,0.583397,0.149603,0.191640
+mpoyraz/wav2vec2-xls-r-300m-cv8-turkish,transformers_ctc,commonvoice_tr,447,0.122774,0.025036,0.146570
+mpoyraz/wav2vec2-xls-r-300m-cv8-turkish,transformers_ctc,issai,453,0.416793,0.096889,0.135329
+mpoyraz/wav2vec2-xls-r-300m-cv8-turkish,transformers_ctc,openslr_tr,160,0.399948,0.096559,0.039539
+openai/whisper-base,openai-whisper,commonvoice_tr,447,0.517901,0.173889,0.076134
+openai/whisper-base,openai-whisper,issai,453,0.567607,0.302055,0.304003
+openai/whisper-base,openai-whisper,openslr_tr,160,0.416564,0.176892,0.069947
+openai/whisper-large-v2,transformers,commonvoice_tr,447,0.140125,0.052663,0.104886
+openai/whisper-large-v2,transformers,issai,453,0.208263,0.109101,0.157161
+openai/whisper-large-v2,transformers,openslr_tr,160,0.277988,0.205973,0.145202
+openai/whisper-large-v3,transformers,commonvoice_tr,447,0.106429,0.035031,0.108919
+openai/whisper-large-v3,transformers,issai,453,0.154161,0.069583,0.156830
+openai/whisper-large-v3,transformers,openslr_tr,160,0.157281,0.094714,0.145426
+openai/whisper-large-v3-turbo,transformers,commonvoice_tr,447,0.228068,0.075274,0.030629
+openai/whisper-large-v3-turbo,transformers,issai,453,0.157752,0.071330,0.038041
+openai/whisper-large-v3-turbo,transformers,openslr_tr,160,0.250647,0.181738,0.030582
+openai/whisper-medium,openai-whisper,commonvoice_tr,447,0.186048,0.075438,0.165166
+openai/whisper-medium,openai-whisper,issai,453,0.253188,0.143137,0.276290
+openai/whisper-medium,openai-whisper,openslr_tr,160,0.183408,0.092229,0.181213
+openai/whisper-small,openai-whisper,commonvoice_tr,447,0.303533,0.119164,0.103580
+openai/whisper-small,openai-whisper,issai,453,0.306197,0.144708,0.139658
+openai/whisper-small,openai-whisper,openslr_tr,160,0.272690,0.136588,0.095569
+openai/whisper-tiny,openai-whisper,commonvoice_tr,447,0.673464,0.325519,0.093877
+openai/whisper-tiny,openai-whisper,issai,453,0.683740,0.347800,0.144045
+openai/whisper-tiny,openai-whisper,openslr_tr,160,0.520328,0.195098,0.047318
+selimc/whisper-large-v3-turbo-turkish,transformers,commonvoice_tr,447,0.123263,0.029597,0.028978
+selimc/whisper-large-v3-turbo-turkish,transformers,issai,453,1.041670,0.721175,0.049470
+selimc/whisper-large-v3-turbo-turkish,transformers,openslr_tr,160,0.272094,0.181679,0.026766
+vincespeed/faster-whisper-large-v3-turbo-turkish,faster-whisper,commonvoice_tr,447,0.120875,0.028630,0.098326
+vincespeed/faster-whisper-large-v3-turbo-turkish,faster-whisper,issai,453,0.239520,0.117997,0.106902
+vincespeed/faster-whisper-large-v3-turbo-turkish,faster-whisper,openslr_tr,160,0.193138,0.098831,0.036546