Kokoro-TTS: A Small Text-to-Speech Model that Once Ranked First on TTS Leaderboards

AIbase基地

Published inAI News · 4 min read · Jan 15, 2025

1.5k

With the rapid development of artificial intelligence, speech synthesis technology is gaining increasing attention. Recently, the latest speech synthesis model named Kokoro was officially released on the Hugging Face platform. This model features 82 million parameters, marking an important milestone in the field of speech synthesis.

Kokoro v0.19 ranked first on the TTS (Text-to-Speech) leaderboard in the weeks leading up to its release, outperforming other models with more parameters. This model achieved results comparable to models like XTTS v2 with 467M parameters and MetaVoice with 1.2B parameters, using less than 100 hours of audio data in a monophonic setup. This achievement indicates that the relationship between the performance of traditional speech synthesis models and their parameters, computational load, and data volume may be more significant than previously expected.

For usage, users only need to run a few lines of code in Google Colab to load the model and voice packages, generating high-quality audio. Currently, Kokoro supports both American English and British English, offering multiple voice packages for users to choose from.

The training process for Kokoro utilized Vast.ai's A100 80GB vRAM instances, which are relatively low-cost to rent, ensuring an efficient training process. The entire model was trained with less than 20 training epochs and under 100 hours of audio data. The Kokoro model was trained using public domain audio data and other open-licensed audio to ensure data compliance.

Despite Kokoro's outstanding performance in speech synthesis, it currently does not support voice cloning due to limitations in its training data and architecture. The main training data is focused on long-form reading and narration rather than dialogue.

Model: https://huggingface.co/hexgrad/Kokoro-82M

Experience: https://huggingface.co/spaces/hexgrad/Kokoro-TTS

Key Highlights:

🌟 Kokoro-82M is a newly released speech synthesis model with 82 million parameters, supporting various voice packages.

🎤 The model excels in the TTS field, having ranked first on the leaderboard and trained with less than 100 hours of audio data.

📊 The training of the Kokoro model utilized open-licensed data to ensure compliance, although some functional limitations still exist.

Hugging Face Adds Handy Feature: One-Click Check for Compatible Models

Hugging Face, a leading open-source AI community platform, has launched a highly anticipated new feature: users can quickly see which machine learning models their computer hardware can run via platform settings. Users simply add their hardware information, such as GPU model, to their Hugging Face profile settings (located at the top right corner: Profile Icon > Settings > Local Apps and Hardware).

DeepSeek-V3-0324 Released: Free for Commercial Use, Runs on Consumer-Grade PCs!

DeepSeek quietly released its latest large language model, DeepSeek-V3-0324, causing a stir in the AI industry. This massive 641GB model appeared on the Hugging Face model hub with almost no prior announcement, continuing the company's understated yet impactful release style. Performance leaps rivaling Claude Sonnet3.5 make this release particularly noteworthy.

DeepSeek-V3-0324 Quietly Released: A Low-Key Upgrade that Ignites the Tech World

On March 24, 2025, without any prior announcement, the Chinese AI research institution DeepSeek released the latest version of its flagship language model, DeepSeek-V3-0324, on the Hugging Face platform. This understated yet powerful update quickly sparked heated discussions within the tech community, with numerous developers and AI enthusiasts sharing their initial experiences and expectations. The following is an in-depth report compiled from feedback within the tech community. I. Mysterious Release: A Quiet Debut with 685 Billion...

CMU Team Introduces Meta Reinforcement Fine-Tuning: A Novel Paradigm for Enhancing Large Language Model Reasoning

Large Language Models (LLMs) are constantly evolving in the field of artificial intelligence. Researchers from Carnegie Mellon University (CMU) and HuggingFace recently introduced a new method called Meta Reinforcement Fine-Tuning (MRT). This method aims to optimize the computational efficiency of LLMs during testing, particularly excelling in solving complex reasoning problems. Studies show that existing LLMs struggle with...

Stop Wasting Time! These 5 AI Search Tools Are Secretly Boosting Others' Productivity

AI technology is changing our learning and work habits in unique ways. As one of the vehicles for AI technology, AI search tools offer users unprecedented convenience. This article explores several efficient AI search tools that not only improve the speed at which we acquire information but also enrich our online experience.

Foreign Media: Manus May Not Be China's Second DeepSeek Moment

Recently, a Chinese AI agent platform called Manus has garnered significant attention online. Since its preview launch last week, the platform has rapidly attracted a large user base, with Hugging Face's Head of Product calling it "the most impressive AI tool I've ever seen". Meanwhile, AI policy researcher Dean Ball described Manus as "the most sophisticated computer AI". Manus's official Discord server amassed...