Show HN: Bark.cpp, fast TTS model for multilingual realistic audio generation

#1

Post by el_pa_b » Wed, Apr 24, 2024, 3:42 PM UTC

Show HN: Bark.cpp, fast TTS model for multilingual realistic audio generation
github.com

Re: Show HN: Bark.cpp, fast TTS model for multilingual realistic audio generation

#2

Post by el_pa_b » Wed, Apr 24, 2024, 3:48 PM UTC

Hello!

I ported Suno AI's Bark text-to-speech model in C/C++ to allow fast, realistic, multilingual audio generation on the CPU.

Generating a 5-second audio with vanilla Bark takes 1 minute on a M1 Pro CPU. Using my port in C++ with ggml, it goes down to 15 seconds.

I aim to bring it down to a second to allow on-device real-time audio generation.

Re: Show HN: Bark.cpp, fast TTS model for multilingual realistic audio generation

#3

Post by jilijeanlouis » Wed, Apr 24, 2024, 4:01 PM UTC

el_pa_b wrote: post #2

Hello! I ported Suno AI's Bark text-to-speech model in C/C++ to allow fast, realistic, multilingual audio generation on the CPU. Generating a 5-second audio with vanilla Bark takes 1 minute on a M1 Pro CPU. Using my port in C++ with ggml, it goes down to 15 seconds. I aim to bring it down to a second to allow on-device real-time audio generation.

Congrats that's pretty impressive !

Re: Show HN: Bark.cpp, fast TTS model for multilingual realistic audio generation

#4

Post by el_pa_b » Wed, Apr 24, 2024, 4:10 PM UTC

jilijeanlouis wrote: post #3

el_pa_b wrote: post #2

Hello! I ported Suno AI's Bark text-to-speech model in C/C++ to allow fast, realistic, multilingual audio generation on the CPU. Generating a 5-second audio with vanilla Bark takes 1 minute on a M1 Pro CPU. Using my port in C++ with ggml, it goes down to 15 seconds. I aim to bring it down to a second to allow on-device real-time audio generation.

Congrats that's pretty impressive !

Thank you!