AI Audio2026
Multilingual, real-time ASR for edge CPUs that uses heterogeneous quantization to reduce model size (4.62→1.58 GB) and lower inference latency. Trades some accuracy for 1.6–2.3× faster inference vs. Whisper.cpp and real-time capability on a few CPU threads, making it suitable for memory- and compute-constrained on-device transcription.
AI Dataset2026
Benchmark for joint speaker diarization and speaker-attributed ASR across all 22 scheduled Indian languages, providing ~108 hours of human-corrected, time-aligned, speaker-attributed transcripts. Includes near-field, far-field and in-the-wild recordings with code-mixing and speaker overlap.
AI Audio2026
Multilingual neural text-to-speech model (0.6B params) with zero-shot voice cloning and a bundled 44.1 kHz codec. Preview release targets 11 recommended languages and aims to deliver near-SOTA quality in a compact checkpoint suited for voice cloning and multilingual TTS prototypes.
AI Audio2026
An end-to-end 11B full-duplex speech model for real-time conversational AI that jointly performs streaming speech understanding and generation, enabling ~450 ms turn-taking, barge‑in and live tool calling in a single unified architecture; research use only.
Retrieves short speech segments from MEG recordings with a compact interpretable neural decoder trained against wav2vec 2.0 embeddings, and maps decoder weights to cortical source space to reveal which acoustic and linguistic features drive retrieval.
Supervises audio reasoning by generating per-sample, audio-grounded rubrics that evolve with model rollouts and serve as reinforcement-learning rewards, improving perception and adaptive multi-step reasoning while avoiding reward saturation.
Generates multi‑speaker speech and environmental audio from textual instructions or a reference clip, supporting zero‑shot voice cloning and detailed scene/specification control. Combines a cleaned, captioned dataset with a VAE-based multimodal generator, reward-conditioned quality control, and staged training to improve expressiveness and multi-audio modeling.
AI Dataset2026
100-hour, single-narrator Egyptian Arabic speech corpus with 15,653 aligned clips at 24 kHz for TTS and ASR fine-tuning; studio-consistent audio, machine-generated undiacritized transcripts, CC BY-NC 4.0 (research/non-commercial use).
AI Model2026
Converts raw ASR transcripts into clean written text: adds punctuation and capitalization, expands spoken numbers/dates/times/currencies/emails, removes fillers and resolves self-corrections. Fine-tuned from Qwen3-0.6B (≈0.6B params), 94.8% token accuracy on a 7,519-case English test set; designed for CPU/edge deployment and deterministic post-processing.
AI Audio2026
Generates and edits speech from natural-language instructions plus optional reference audio, supporting zero-shot TTS, content/acoustic/paralinguistic edits, enhancement, and source separation. Open-source 1.5B-parameter base model with a 4-step distilled AuK‑Flash for faster inference.
AI Audio2026
Zero-shot multilingual text-to-speech checkpoint for speech generation and voice cloning with a compact footprint. Features an ~170M-parameter main model plus a bundled ~120M-parameter codec decoder, with primary support for Chinese and English; other languages show more variable quality and long/noisy references reduce fidelity.
AI Model2026
Open-weights LLM fine-tuned for phone-based voice agents that prioritizes low latency and reliable tool/function calling. Based on NVIDIA Nemotron 3 Nano (30B total, 3.5B active), supports very long contexts (262,144 tokens) and recommends temperature=0 with thinking disabled for deployment.