NeuTTS Air is the world's first ultra-realistic, edge-side text-to-speech (TTS) language model with instant voice cloning capabilities. Built on a large language model backbone with 0.5B parameters, it brings natural voice, real-time performance, built-in security, and speaker cloning capabilities to local devices.

Audio ProcessingGguf

NeuTTS Air is the world's first ultra-realistic on-device text-to-speech model with instant voice cloning capabilities. It is built on a large language model backbone with 0.5B parameters and can achieve natural speech generation, real-time performance, and speaker cloning on local devices.

Audio ProcessingGguf

NeuTTS Air is the world's first ultra-realistic, device-side text-to-speech (TTS) language model with instant voice cloning capabilities. Built on a 0.5B large language model backbone network, it can bring natural voice, real-time performance, built-in security features, and speaker cloning capabilities to local devices.

Audio ProcessingSafetensors

SE-DiCoW is a target speaker multi-speaker automatic speech recognition model developed jointly by BUT Speech@FIT, JHU CLSP/HLTCOE, and CMU LTI. Based on Whisper large-v3-turbo, this model significantly improves the recognition accuracy in highly overlapped multi-speaker scenarios through a self-registration mechanism and improved data augmentation techniques.

Audio ProcessingTransformersOther

Smart Turn v2 is an open-source semantic voice activity detection (VAD) model that determines whether the speaker has finished speaking by analyzing the raw waveform.

Audio ProcessingSafetensorsOther

A state-of-the-art speaker segmentation model optimized for the Apple Neural Engine, capable of real-time speaker separation on the device, supporting any language, and trained with acoustic features, with research-level performance.

Audio ProcessingCoreml

SAUTE is a lightweight Transformer architecture with speaker perception ability, designed for effectively modeling multi-speaker dialogues.

Natural Language ProcessingTransformersEnglish

This model is built on DeBERTa-v3-large and can predict the gender of anonymous speakers or authors based on the content of English texts.

Natural Language ProcessingTransformersEnglish

Provide neural text-to-speech synthesis for Bambara, supporting 10 native Bambara speakers and generating high-fidelity audio.

Audio ProcessingTransformersOther

Bielik-11B-v2.6-Instruct is a large Polish language model developed by SpeakLeash and ACK Cyfronet AGH, fine-tuned based on Bielik-11B-v2, suitable for instruction following tasks.

Natural Language ProcessingTransformerspolish

A speech emotion classification model fine-tuned based on facebook/wav2vec2-base-960h, specifically designed to accurately identify and classify 8 different emotional states of speakers from audio signals.

Audio ProcessingTransformersEnglish

An audio classification model based on WavLM-Large, used for accurately identifying the age and gender of speakers.

Audio ProcessingSafetensorsEnglish

The quantized version of Bielik-4.5B-v3.0-Instruct, based on a large Polish instruction fine-tuned language model with 4.5 billion parameters, optimized for efficient inference and deployment, and built with the concept of 'making knowledge accessible to everyone'.

Natural Language ProcessingGguf

Bielik-4.5B-v3 is a Polish generative text model with 4.6 billion parameters, developed by SpeakLeash in collaboration with ACK Cyfronet AGH, trained on a curated Polish corpus.

Natural Language ProcessingTransformersOther

This model is the FP8 quantized version of Bielik-4.5B-v3.0-Instruct, utilizing AutoFP8 technology to quantize weights and activations into FP8 data type, reducing approximately 50% of disk space and GPU memory requirements.

Natural Language ProcessingGgufOther

This is an FP8 dynamic quantization version based on the Bielik-1.5B-v3.0-Instruct model, adapted for vLLM or SGLang inference frameworks. It uses AutoFP8 quantization technology to reduce parameter bytes from 16-bit to 8-bit, significantly lowering disk space and GPU VRAM requirements.

Natural Language ProcessingGgufOther

This is a 1.5B parameter instruction fine-tuned model for Polish, developed based on the SpeakLeash Bielik series, suitable for text generation tasks.

Natural Language ProcessingGgufOther

Bielik-4.5B-v3.0-Instruct-GGUF is a Polish large language model released by SpeakLeash, converted from Bielik-4.5B-v3.0-Instruct to GGUF quantized format, suitable for local inference.

Natural Language ProcessingGgufOther

Bielik-4.5B-v3-Instruct is a 4.6 billion parameter Polish generative text model, fine-tuned based on Bielik-4.5B-v3, demonstrating exceptional Polish language comprehension and processing capabilities.

Natural Language ProcessingTransformersOther

Bielik-1.5B-v3-Instruct is a 1.6 billion parameter Polish generative text model, fine-tuned for instructions based on Bielik-1.5B-v3, developed by SpeakLeash in collaboration with ACK Cyfronet AGH.

Natural Language ProcessingTransformersOther