5 repositorios
Processes that convert model samples into final audio waveforms with configurable formats.
Distinct from Audio Generation Models: Focuses on the reconstruction of waveforms from model samples
Explore 5 awesome GitHub repositories matching artificial intelligence & ml · Audio Sample Reconstruction. Refine with filters or upvote what's useful.
Audiocraft is a deep learning audio library and machine learning framework designed for training, fine-tuning, and evaluating generative models for music and sound effects. It functions as a text-to-music generative model and a neural audio codec, providing the tools necessary to compress audio signals into discrete representations and synthesize high-fidelity waveforms from textual descriptions. The framework is distinguished by its ability to combine multiple conditioning signals, allowing for the generation of audio based on text prompts, melodic excerpts, or style-based audio clips. It al
Generates audio outputs from model samples using configurable batch sizes and output formats.
Amphion is an audio generation toolkit designed for the research and development of models that synthesize speech, music, and environmental sound effects. It provides a standardized framework for reproducible audio synthesis, incorporating a text-to-speech engine and a voice conversion framework. The project specializes in transforming audio identities, allowing for the modification of speaker accents and voice identities while preserving original rhythm and style. It also includes capabilities for singing voice synthesis and the generation of environmental soundscapes from text descriptions
Produces high-quality audio waveforms from intermediate representations using specialized neural vocoders.
Lyra es un framework de compresión de voz y códec de voz de baja tasa de bits diseñado para transmitir audio de alta calidad a través de redes con ancho de banda limitado. Utiliza un códec de audio de tasa de bits adaptativa para equilibrar la calidad del audio y el ancho de banda de la red durante sesiones activas. El proyecto emplea compresión de audio generativa, utilizando redes neuronales para sintetizar señales de voz a partir de datos mínimos y reconstruir detalles de audio faltantes. Esto permite la reconstrucción de voz de alta calidad a partir de flujos de bytes altamente comprimidos. El sistema cubre voz sobre IP optimizada para ancho de banda y comunicación de voz en tiempo real, centrándose en la compresión de voz de baja tasa de bits para mantener la estabilidad de las llamadas. Sus capacidades incluyen el ajuste dinámico de la tasa de bits de audio y el procesamiento de voz para evitar retrasos y caídas de señal en entornos de red inestables.
Uses generative models to reconstruct high-quality speech waveforms from model samples.
EnCodec is a neural audio codec and compression tool designed to transform raw audio waveforms into discrete codes and reconstruct them back into sound. It functions as a system for neural audio representation, converting continuous audio signals into sequences of integer indices for use in generative AI tasks. The project utilizes a residual vector quantizer, which employs multiple layers of codebooks to represent audio signals with high precision at low bitrates. This approach allows the system to compress audio to discrete codes and perform low bitrate audio coding for efficient transmissi
Converts discrete compressed codes back into playable audio signals without distortion.
Neural Amp Modeler is an open-source project that captures the tonal character of analog audio gear by training a neural network on paired dry and reamped audio recordings. It provides a complete pipeline for learning how a guitar amplifier, effects pedal, or other audio device transforms a signal, then exports the trained model into a portable file format for use in other applications. The project centers on a file-format-based approach to model distribution, where each trained neural network is saved as a single .nam file that can be shared and loaded by different host applications. A real-
Trains neural networks from paired dry and reamped audio recordings to learn gear transformations.