官网：KittenML/KittenTTS: State-of-the-art TTS model under 25MB 😻
对了，这个模型最大的问题是不会中文。
Kitten TTS介绍
Kitten TTS是一个开源的现实文本到语音模型，只有1500万个参数，专为轻量级部署和高质量语音合成而设计。目前处于开发者预览阶段
✨ 特性
超轻：型号大小小于25MB
CPU优化：在任何设备上无需GPU即可运行
高品质语音：提供多种高级语音选项
快速推理：针对实时语音合成进行了优化
安装Kitten TTS
直接使用pip安装即可。
pip install https://github.com/KittenML/KittenTTS/releases/download/0.1/kittentts-0.1.0-py3-none-any.whl使用Kitten TTS
直接在python里调用即可
from kittentts import KittenTTS
m = KittenTTS("KittenML/kitten-tts-nano-0.2")
audio = m.generate("This high quality TTS model works without a GPU", voice='expr-voice-2-f' )
# available_voices : [ 'expr-voice-2-m', 'expr-voice-2-f', 'expr-voice-3-m', 'expr-voice-3-f', 'expr-voice-4-m', 'expr-voice-4-f', 'expr-voice-5-m', 'expr-voice-5-f' ]
# Save the audio
import soundfile as sf
sf.write('output.wav', audio, 24000)整体输出
>>> from kittentts import KittenTTS
>>> m = KittenTTS("KittenML/kitten-tts-nano-0.2")
config.json: 177B [00:00, 34.1kB/s]
kitten_tts_nano_v0_2.onnx: 100%|████████████████████████████████████████████████████| 23.8M/23.8M [01:20<00:00, 295kB/s]
voices.npz: 100%|██████████████████████████████████████████████████████████████████| 10.3k/10.3k [00:00<00:00, 13.6MB/s]
>>> import soundfile as sf
>>> audio = m.generate("This high quality TTS model works without a GPU", voice='expr-voice-2-f' )
>>> sf.write('output.wav', audio, 24000)
>>> m = KittenTTS("KittenML/kitten-tts-nano-0.2")
KeyboardInterrupt
>>> audio = m.generate("现在，当用户在创建生字本时，输入的汉字会自动从3500个常用汉字数据库中获取信息", voice='expr-voice
-2-f' )
2025-12-21 00:13:30.158456312 [E:onnxruntime:, sequential_executor.cc:572 ExecuteKernel] Non-zero status code returned while running Expand node. Name:'/bert/Expand' Status Message: invalid expand shape
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/skywalk/.local/lib/python3.10/site-packages/kittentts/get_model.py", line 37, in generate
return self.model.generate(text, voice=voice, speed=speed)
File "/home/skywalk/.local/lib/python3.10/site-packages/kittentts/onnx_model.py", line 102, in generate
outputs = self.session.run(None, onnx_inputs)
File "/home/skywalk/.local/lib/python3.10/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 287, in run
return self._sess.run(output_names, input_feed, run_options)
onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Non-zero status code returned while running Expand node. Name:'/bert/Expand' Status Message: invalid expand shape
>>> audio = m.generate("常用汉字数据库中获取信息", voice='expr-voice-2-f' )
>>> sf.write('output1.wav', audio, 24000)如果是汉字信息，它会一直用英文说：chinese letter chinese letter 。所以它是不会说中文的。
总结
使用效果很不错，唯一遗憾就是目前不支持中文！
所以现在国内应该用不着。
调试
import kittentts报错
File "E:\py312\Lib\site-packages\pydantic\v1\typing.py", line 529, in update_field_forward_refs
update_field_forward_refs(sub_f, globalns=globalns, localns=localns)
File "E:\py312\Lib\site-packages\pydantic\v1\typing.py", line 520, in update_field_forward_refs
field.type_ = evaluate_forwardref(field.type_, globalns, localns or None)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "E:\py312\Lib\site-packages\pydantic\v1\typing.py", line 66, in evaluate_forwardref
return cast(Any, type_)._evaluate(globalns, localns, set())
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: ForwardRef._evaluate() missing 1 required keyword-only argument: 'recursive_guard'根据报错信息和相关库的特性，问题核心在于 pydantic 库在处理类型引用时缺少关键参数 recursive_guard。以下是具体分析和解决方案：
核心问题分析
-
库依赖冲突：
KittenTTS依赖spacy，而spacy依赖pydantic。pydantic版本 2.7.1 在 Python 3.13.4 中触发了ForwardRef._evaluate()参数缺失错误（）。
-
关键库版本：
pydantic2.7.1：存在recursive_guard参数缺失的 bug。spacy3.5.0+：依赖pydantic2.x，但未修复该 bug。
解决方案
1. 降级 pydantic 版本
pip install pydantic==1.10.12升级 spacy 到最新版
pip install spacy --upgrade做了这两步，还是没解决。
换一台ubuntu22.04 解决了。
报错huggingface_hub.errors.LocalEntryNotFoundError: An error happened while trying to locate the file on the Hub and we cannot find the requested files in the local cache
File "/home/skywalk/.local/lib/python3.10/site-packages/httpx/_transports/default.py", line 118, in map_httpcore_exceptions
raise mapped_exc(message) from exc
httpx.ConnectError: [Errno 93] Protocol not supported
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/skywalk/.local/lib/python3.10/site-packages/kittentts/get_model.py", line 24, in __init__
self.model = download_from_huggingface(repo_id=repo_id, cache_dir=cache_dir)
File "/home/skywalk/.local/lib/python3.10/site-packages/kittentts/get_model.py", line 68, in download_from_huggingface
config_path = hf_hub_download(
File "/home/skywalk/.local/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 89, in _inner_fn
return fn(*args, **kwargs)
File "/home/skywalk/.local/lib/python3.10/site-packages/huggingface_hub/file_download.py", line 1007, in hf_hub_download
return _hf_hub_download_to_cache_dir(
File "/home/skywalk/.local/lib/python3.10/site-packages/huggingface_hub/file_download.py", line 1135, in _hf_hub_download_to_cache_dir
_raise_on_head_call_error(head_call_error, force_download, local_files_only)
File "/home/skywalk/.local/lib/python3.10/site-packages/huggingface_hub/file_download.py", line 1734, in _raise_on_head_call_error
raise LocalEntryNotFoundError(
huggingface_hub.errors.LocalEntryNotFoundError: An error happened while trying to locate the file on the Hub and we cannot find the requested files in the local cache. Please check your connection and try again or make sure your Internet connection is on.设定hf镜像加速
export HF_ENDPOINT=https://hf-mirror.comok了