Subscribe
Sign in
Adobe Podcast Enhance Speech v2 — AI to make spoken audio sound professional
Enhance Speech v2 is an advanced AI tool that improves audio quality by eliminating background noise and enhancing voice clarity. It makes recordings sound professional, making it ideal for anyone looking to enhance their audio quality.
30
Replies
@markcwebster Hey Mark, v1 has been one of the most useful tool for me and my team. What makes this v2 better over v1?
This looks exciting - especially b/c I often have a less-than-ideal audio setup. Are there specific enhancements or features in V2 that make it particularly effective for improving conversational audio with challenging background noise?
@paddystobbs Hey Paddy. There's two main changes in v2:
-A completely new AI model that handles a much broader range of audio issues, resulting in more natural sounding speech.
-v2 also separates out any background noise from the audio, and with Premium, let's you mix back in just the right amount when needed.
Give it a try, let us know what you think...
Well done team! Just realeased a video putting it through the paces. TLDW; It's good and you can really trust the audio output and quality over the v1 that sometimes suffered the odd robotic glitches and some EQ changes.
I've successfully used this for editing out some bad echo in my last YouTube video. Highly recommended! The fact that it's free (I have an Adobe account since forever) is even better. Great work to everyone involved in its development!
@samalyx Hey Samar! Here's a high level explanation: Enhance Speech is trained on a large dataset of professionally recorded spoken audio, basically teaching the AI "this is what professional sounding audio sounds like." When a file is uploaded to the service, the AI is asked "based on what you know about the specific characteristics of professionally recorded audio, what would this file sound like had it been professionally recorded?" and it returns the enhance audio track. We're not generating any synthetic voices or anything, it's all just working with the characteristics of the audio itself. It's unique based on both the approach the model takes, and the largely proprietary dataset we used for the training. Hope that helps!
"Huge congrats to the Adobe team on the launch of Enhance Speech v2 today! Quick question: Can Enhance Speech v2 also handle accent neutralization to some extent, or is that on the roadmap for future updates?"
@ihuzaifashoukat Hey Huzaifa. Enhance Speech is focused on improving the quality of the spoken audio, to make it sound as it would have in a professional recording studio. We don't alter the accents in any way, and have no plans to in the future. Hope that helps!
I would recommend this tool to anyone who works with audio content and wants their recordings to sound as clean and professional as possible without too much effort.
I’m already a fan of the V1 speech enhancement for reels' voiceovers, and I’m excited to give V2 a try!
Congrats on the launch, wishing you lots of success!
Congratulations on the launch, Adobe Podcast team! How does Adobe Podcast Enhance Speech v2 handle recordings with varying voice tones, and does it allow for customizable adjustments for different audio environments?
Adobe Podcast Enhance Speech v2 looks like a game-changer for content creators! The ability to remove background noise and enhance voice clarity effortlessly is impressive. I’m curious does this tool support real-time audio processing, or is it mainly for post-production? Amazing work by the team!
Ok first off I'm a huge fan and long-time user of Adobe Enhance. Big ups to the whole team who developed this algorithm. However, since the V2 rollout I noticed that the 'Strength' feature no longer works. At 0% it is disabled, but as soon as you turn it up to 1%, it is turned on fully, functioning at 100%. My workaround has been to just download the 100% file and layer with the raw file accordingly, but would very much appreciate a bug fix here!! Thanks y'all
@andy_berger_awm Hey Andy, thanks for the feedback. The slider for v2 works a bit differently. Since v2 separates out the background/reverb, the slider now allows you to mix that background audio back into the enhanced speech track. If the audio was pretty clean to begin with, it may seem like the slider isn't doing anything.
BUT we've gotten some feedback similar to yours that it can be confusing on certain files, and we may make some tweaks to how it works in the future. Appreciate your thoughts, super helpful towards making Enhance Speech even better.
V2 is AMAZING! I literally had construction outside my window while recording my podcast (Im a newbie podcaster so cant afford so the full studio set up or to rent a studio) and I was stuck for 2 months. V1 made the voices robotic) but thanks to v2 I can now publish. I'm a believer.
OH I should mention the type of sounds were "banging" noises which are MUCH harder to remove.
What a great tool! I actually used this yesterday for a recording in my home office with my iPhone. I live in a city area so the outside noise is fairly prevalent, but this tech crushed it! The output sounded like I was using a $1,000 microphone in a recording studio. Still can't believe how good it sounds.
My only minor suggestion would be to have the ability to upload a movie file but be able to download just the audio file produced. Thanks!
Adobe Podcast