SuperAPI AI Directory
Partner中文AI newsFavorites

All AI tools

35,738 results · Page 390 of 993

Soul Cinema

AI images

Soul Cinema is a cinematic AI image generator: users simply describe the shot in words, and the system returns still frames with a cinematic feel: including cinematographer-level lighting, a consistent cast, a reusable color grading style, and full-resolution export. Its signature features include Soul ID (locking the same face and wardrobe across a sequence of up to six frames to maintain character consistency), palette lock (preserving specified hexadecimal color values so colors do not drift on re-render), and support for export at up to 8K resolution with no watermark, with files output exactly as produced by the model and not re-encoded. Soul Cinema runs entirely in the browser; after signing in with Google and selecting a paid plan, it is ready to use. It has no free tier, and all plans are paid, offering different tiers of 120, 600, and 2000 credits per month, and can be canceled at any time.

Read more

Soundora

AI audio

Soundora is an AI music generator that creates songs and instrumental works based on lyrics or creative descriptions provided by users. Users can add a title and lyrics, generate lyrics with AI, choose a genre and vocal direction, and compare different song versions within the workspace. Its AI music generator uses Suno V6 capabilities and supports both vocal and vocal-less instrumental creation. It is suitable for a variety of scenarios such as video soundtracks, game soundtracks, podcasts, advertisements, courses, and fitness training. In terms of pricing, the credit price is displayed before each operation is submitted, and some accounts can be configured with a daily free quota; copyright and commercial licensing depend on the input content, the generation service provider, and the plan used, and the applicable licensing terms must be reviewed before publishing or commercial use.

Read more

Spark-TTS

AI audio

Spark-TTS is an efficient text-to-speech (TTS) system based on a large language model (LLM), developed by the SparkAudio team, with official PyTorch inference code provided. It is built entirely on Qwen2.5, requiring no additional generative models such as flow matching, and directly reconstructs audio from codes predicted by the LLM, thereby simplifying the pipeline and improving efficiency. The model supports zero-shot voice cloning, reproducing a speaker's timbre without training on a specific voice, and is suitable for cross-lingual and code-switching scenarios. It also supports both Chinese and English, and can synthesize multilingual speech with high naturalness and accuracy. In addition, Spark-TTS supports creating virtual speakers by adjusting parameters such as gender, pitch, and speaking rate. The project provides detailed installation steps (Linux and Windows guides), demo scripts, a Web UI interface, and an Nvidia Triton inference service deployment solution. The official website emphasizes that the model is intended only for academic research, education, and lawful applications, and strictly prohibits unauthorized voice cloning, impersonation, fraud, and other illegal acts.

Read more