I was working on a local/offline speech to text app and needed to figure out a way to use Whisper. There was one constraint, I cannot have any additional dependency. I needed to minimize the app's disk and runtime footprint to be able to make it useful in a daily usage.
After moving forward on this front on my own, I landed me on whisper.cpp, an Open Source project created by Georgi Gerganov (he also created the popular llama.cpp). Surprised to see that it already has a massive interest (34k stars) and has received massive contribution from close to 400 other people.
My experiments continued and this review is based on what I have learned after trying Whisper.cpp for a day.
I was working on a local/offline speech to text app and needed to figure out a way to use Whisper. There was one constraint, I cannot have any additional dependency. I needed to minimize the app's disk and runtime footprint to be able to make it useful in a daily usage.
After moving forward on this front on my own, I landed me on whisper.cpp, an Open Source project created by Georgi Gerganov (he also created the popular llama.cpp). Surprised to see that it already has a massive interest (34k stars) and has received massive contribution from close to 400 other people.
My experiments continued and this review is based on what I have learned after trying Whisper.cpp for a day.
By now, most of you'd already know about Whisper. If not, or just want to dive deeper, check out following links
* Paper - https://cdn.openai.com/papers/whisper.pdf
* Whisper intro by OpenAI - https://github.com/openai/whisper
* Official Whisper python package by OpenAI - https://github.com/openai/whisper