---
language:
  - tr
  - zh
  - en
  - ja
  - es
  - ar
license: other
license_name: bilibili-model-license
license_link: LICENSE
library_name: indextts
pipeline_tag: text-to-speech
tags:
  - text-to-speech
  - tts
  - zero-shot
  - voice-cloning
  - multilingual
  - cross-lingual
  - lora
  - finetune
  - turkish
base_model: IndexTeam/IndexTTS-2.5
base_model_relation: finetune
datasets:
  - freyavoice/freya-tr-eval
---

# IndexTTS-2.5 Turkish (v2)

Fine-tuned GPT weights that add **Turkish** to [IndexTeam/IndexTTS-2.5](https://huggingface.co/IndexTeam/IndexTTS-2.5) with zero-shot voice cloning. The five base languages (zh, en, ja, es, ar) are kept. Trained with [index-tts-2.5-ft](https://github.com/oberkayb/index-tts-2.5-ft).

**v2 cuts word errors 9×: Freya-TR-Eval WER drops from 12.36% (v1) to 1.32%.** v1 swapped short Turkish words for other words (*yağmur → "arkadaş mur"*). The cause was text-embedding rows the base model had never trained. v2 trains exactly those rows.

## Freya-TR-Eval

[Freya-TR-Eval](https://huggingface.co/datasets/freyavoice/freya-tr-eval) has 495 everyday Turkish sentences. We ran it with the protocol of the [PatientDesk re-run](https://huggingface.co/datasets/PatientdeskAI/freya-tr-eval-rerun): three seeded takes per sentence, an 8 kHz band, faster-whisper large-v3 (beam 5), EMA's `score_norm` and UTMOS22-strong. The voice reference is the same one the cloning systems used there (Anka TTS `male`).

| System | WER % | CER % | Clean | UTMOS |
|---|---|---|---|---|
| EMA Lightning | 1.04 ± 0.17 | 0.22 | 93.7% | 3.30 |
| Trendyol-TTS | 1.16 ± 0.16 | 0.23 | 93.1% | 3.82 |
| Gemini 3.8 Flash TTS | 1.23 ± 0.10 | 0.29 | 92.9% | 3.53 |
| **IndexTTS-2.5 Turkish v2 (this model)** | **1.32 ± 0.10** | **0.26** | **92.2%** | **3.17** |
| Gemini 3.8 Flash-Lite TTS | 1.34 ± 0.19 | 0.27 | 91.9% | 3.63 |
| Anka TTS | 1.42 ± 0.06 | 0.30 | 91.6% | 3.52 |
| ElevenLabs v4 | 1.43 ± 0.12 | 0.27 | – | – |
| Chatterbox Multilingual | 1.91 ± 0.35 | 0.43 | 88.8% | 3.51 |
| XTTS-v2 | 3.67 ± 0.37 | 1.50 | 77.1% | 2.97 |
| FreyaTTS-small | 12.02 ± 0.00 | 4.95 | 47.1% | 2.81 |
| IndexTTS-2.5 Turkish v1 | 12.36 ± 0.43 | 5.72 | 43.5% | 3.08 |

- **Our rows** (bold and v1) were run by us on 2026-10-08. All other rows are quoted from the PatientDesk re-run of 2026-10-07; it lists 19 systems.
- **Clean** is the share of sentences with no word error. v2 reads 427 of the 495 sentences correctly in all three takes; v1 does so for 161.
- **Speed:** RTF is 0.83 on an RTX 5070 Ti (bf16, `num_beams=3`). This is not comparable with the L4 numbers in the re-run.
- **Raw data:** transcripts, UTMOS per clip and timing are in [`benchmark/`](./benchmark).

## v1 vs v2: listen

Same sentence, reference voice and seed (take 0). The v1 column shows what Whisper heard.

<table>
<tr><th>Sentence</th><th>v1 (Whisper heard)</th><th>v1</th><th>v2</th></tr>
<tr><td>Yağmur birden bastırınca hepimiz sırılsıklam olduk!</td><td><i>Arkadaş mur</i> birden bastırınca hepimiz <i>kırılsıklam</i> olduk.</td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0129_v1.wav"></audio></td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0129_v2.wav"></audio></td></tr>
<tr><td>Ne demek istediğini bilmiyorum.</td><td>Ne <i>biraz geldiğimi zatım</i>.</td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0320_v1.wav"></audio></td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0320_v2.wav"></audio></td></tr>
<tr><td>Akşamüstü kuaföre uğrayıp saçlarımı biraz kestirmek istiyorum.</td><td><i>Akşam düşüğü</i> kuaföre uğrayıp <i>hasırlarımı</i> biraz kestirmek istiyorum.</td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0232_v1.wav"></audio></td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0232_v2.wav"></audio></td></tr>
<tr><td>Dünkü yemekten kalanları ısıtıp öğle yemeğinde afiyetle yedik.</td><td><i>Dört geçten</i> kalanları <i>sırtıp</i> öğle yemeğinde <i>afasırla</i> yedik.</td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0108_v1.wav"></audio></td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0108_v2.wav"></audio></td></tr>
<tr><td>Birlikte sinemaya gitmeyeli neredeyse iki ay olmuş.</td><td><i>Artık</i> sinemaya gitmeyeli <i>söyleyse 2 ayıdır</i>.</td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0083_v1.wav"></audio></td>
<td><audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/0083_v2.wav"></audio></td></tr>
</table>

Reference voice: <audio controls src="https://huggingface.co/berrkkaayy/IndexTTS-2.5-Turkish-LoRA/resolve/main/samples/reference_anka_male.wav"></audio>

The reference is the AI-generated `male` voice of [Anka TTS](https://pypi.org/project/anka-tts/) (not a real person), licensed CC-BY-NC-4.0. The clips in `samples/` are derived from it and carry the same license.

## Files

| File | Description |
|---|---|
| `gpt.pth` | **v2**, step 13000. Merged LoRA export; drop-in replacement for `checkpoints/gpt.pth`. |
| `export_metadata.json` | v2 provenance and training settings |
| `v1/gpt.pth`, `v1/export_metadata.json` | Previous release (step 29000, 2026-08-19), kept for comparison |
| `samples/`, `benchmark/` | Listening samples and Freya-TR-Eval raw results |

All other weights (`s2mel.pth`, codec, vocoder, tokenizer, …) come from [IndexTeam/IndexTTS-2.5](https://huggingface.co/IndexTeam/IndexTTS-2.5).

## Quick start

```bash
git clone https://github.com/index-tts/index-tts.git && cd index-tts
uv sync --all-extras
hf download IndexTeam/IndexTTS-2.5 --local-dir=checkpoints
hf download berrkkaayy/IndexTTS-2.5-Turkish-LoRA gpt.pth --local-dir=checkpoints/tr

cd .. && git clone https://github.com/oberkayb/index-tts-2.5-ft.git && cd index-tts-2.5-ft
uv run --project ../index-tts python scripts/synthesize.py \
  --gpt-checkpoint ../index-tts/checkpoints/tr/gpt.pth \
  --lang tr --normalizer turkish --case tr_lower \
  --prompt-audio ref.wav --text "Merhaba, bugün hava çok güzel." --output out/tr.wav
```

Turkish needs the **itts25ft text frontend** (`turkish` normalizer + `tr_lower`). Training and inference must match. If you call `IndexTTS2.infer` directly, clean the text with the frontend and pass `lang="tr", text_normalization=False`. The fine-tuning repo also ships a WebUI (`webui.bat`) and an MCP server (`mcp_server.py`).

## Training (v2)

| Item | Value |
|---|---|
| Data | 486.5 h Turkish, 143,609 utterances, 3,652 speakers; 286,606 prompt/target pairs |
| Validation | 674 pairs from 20 held-out speakers; step 13000 had the best validation loss |
| Method | LoRA r32/α64 on all GPT blocks, `lang_embedding` row 9 (seeded from `es`), mel/text heads |
| New in v2 | **1,530 untrained text-embedding rows** trained (`--train-text-rows`). These rows carry 37% of the Turkish token occurrences. Gradient-masked, LR ×10, no weight decay, seeded from trained sub-word pieces. |
| Data quality | ASR transcripts filtered by duration, chars/s and repeated text; clips over 25 s dropped, not truncated |
| Optimisation | LR 1e-4, bf16, effective batch 32, 13,000 steps (~1.45 epochs) on an RTX 5070 Ti + RTX 4070 |
| Frozen | Emotion branch, speaker projection, all text rows the base model already trained |

## Limitations

- Turkish requires the itts25ft frontend; stock `webui.py` does not apply it.
- Numbers are best written out as words. Long text is segmented, and prosody may reset between segments.
- UTMOS is an English-trained proxy and depends on the reference voice.
- Use is subject to the [bilibili Model Use License Agreement](./LICENSE) inherited from IndexTTS-2.5.

## Türkçe özet

**v2**, Freya-TR-Eval'de kelime hata oranını **%12,36'dan %1,32'ye** indirdi; doğru okunan cümle oranı %43,5'ten %92,2'ye çıktı. v1'deki kelime karışmasının sebebi, taban modelin Türkçe token satırlarını hiç eğitmemiş olmasıydı. v2 tam olarak bu 1.530 satırı eğitiyor. `gpt.pth` artık v2. Eski sürüm `v1/` klasöründe duruyor. Kullanım için yukarıdaki *Quick start* bölümüne bak; Türkçe metin işleme (`turkish` + `tr_lower`) şart.

## Citation

```bibtex
@misc{li2026indextts25technicalreport,
  title={IndexTTS 2.5 Technical Report},
  author={Yunpei Li and others},
  year={2026},
  eprint={2601.03888},
  archivePrefix={arXiv},
}
@misc{pamuk2026freyattstechnicalreport,
  title={FreyaTTS Technical Report},
  author={Ahmet Erdem Pamuk and Ömer Yentür and Ahmet Tunga Bayrak and Yavuz Alp Sencer Öztürk and Mustafa Yavuz},
  year={2026},
  eprint={2607.09530},
  archivePrefix={arXiv},
}
```