Launched this week
Flow Type
Just dictate, Flow type has been re-write it.
11 followers
Just dictate, Flow type has been re-write it.
11 followers
LLM powered voice transcription and content enhancement - sasisprite/speech-to-text
11 followers
11 followers
The voice transcription is impressively quick and it cleaned up my rambling notes into something actually coherent. Curious to see how it handles heavier accents or noisier recordings.
@sagldanefyuv Thanks for the feedback, It runs on the extra-large OpenAI Whisper model, allowing you to simply whisper to it. Additionally, it can clean up noisy backgrounds, helping you achieve a clean outcome.
Finally tried a voice transcription tool that didn't butcher my technical terms. The LLM cleanup on the backend makes a real difference compared to raw Whisper output.
@sudenuriskr5xt Thank you for your feedback. We are working on an adaptive LLM cleanup. Based on your continuous engagements and the current tool you are using, we plan to rewrite the LLM cleanup to better suit your needs. This will enable you to place content as desired, whether it's for a messaging tool, code editing tool, or emails. We intend to add these features to enhance the cleanup process.
Voice transcription feels pretty snappy and the LLM cleanup actually tightens my rambling into something readable. Curious to see how it handles really noisy audio next.
@ramazanszglroz Flowtype is powered by the extra-large OneOpenAI Whisper model, which helps to clean background noises and even allows you to whisper to it. It also supports low-quality audio, such as 16kHz audio.
A nice clean tool for voice transcription. One thing I'd love is the ability to set custom vocab or industry terms so the LLM enhancement doesn't mangle jargon or product names. Even a small list users can save per project would make the output way more reliable for technical work.
@nkurumak28108 I 100% agree on this. We will see how we can add this to our next release. Thanks for all the feedback.