@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 1/ Ugh, my damn foggy narcolepsy brain and not looking at your profile info :P. My bad on that. Apologies to Ian as well; he's just my go-to capeless hero for this kind of stuff.
Wall 'o text incoming...
Wall 'o text incoming...
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 2/ 1. Separate Audio Channels Makes sense to a point. FTL uses Opus, which supports 255 audio channels, and audio coupling (could pair all multi-channel game audio, and voice audio separately).
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 3/ RTMP can also support multi-channel audio via streaming multiple mono channels to the ingest server (and even supports text streams, which Twitch/YouTube use for CC data). Most streamers already have their audio split into different channels (especially for recording).
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 4/ I know I have 4 myself. I couldn't find any concrete technical details on the FTL protocol (and wasn't about to read through the SDK's source code) regarding support for text streams, but in the end it's MS' tech so it could be added if you wanted to.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 5/ That still begs the question though, why would you want to limit CC to what the streamer is saying? CC typically includes both dialogue and audio descriptors. Television broadcasts typically also include the option to display video descriptors as CC (Azure Video Indexer).
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 6/ This would go towards benefiting both those that are hard of hearing, as well as the visually impaired. There are countless other benefits for supporting multi-channel audio.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 7/ I've always wanted the ability for viewers to mute my music audio channel if they don't want to listen to it, or even my discord / sound effects audio.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 8/ Like I said, I'm already splitting these sources on my end, so it's the streaming platform itself that's the blocker in this situation. Even if all of the above were untrue, and you only had one audio stream to work with, Azure's Cognitive Services already has
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 9/ 'Speaker Recognition' / 'Conversation Transcription (diarization)' for distinguishing between different speakers based on their unique voice signature. The ML/neural-net tech is already there on MS' side, and the data sources are already being provided from the broadcasters.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 10/ 2. Cost Efficiency I'm kind of puzzled by this concern, as you're already passing all of the cost onto end users, and even encouraging it with posts like this.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 11/ Most people I've seen that implement CC are utilizing a Cloud AI provider service to do so, with the only other option being the utilization of something like Windows' (inferior) speech recognition.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 12/ There's also the issue of implementing CC being a large technical hurdle for most streamers, thus drastically limiting those that will implement it.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 13/ There's also a large percentage that simply don't see the benefit to doing so, or are oblivious to the fact that CC is possible or that there's a viewer demographic that requires it. IMO, this should be a non-issue and should be considered an operating cost by Mixer.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 14/ It's one thing to charge Azure customers for the use of Cognitive Services for their own business/personal uses, but we're talking about your own platform here, which is already utilizing Azure's infrastructure. By implementing this yourselves, you add groundbreaking…
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 15/ …accessibility features (Good PR, increased viewership), you increase awareness of accessibility needs and the advantage of utilizing your cloud services to do so, and countless other possibilities for enhancing the platform.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 16/ It would be a no-brainer for me in Mixer's situation, to 'drink my own champagne' and lead the charge on automating this entire process at no charge to broadcasters.
@LadieAuPair @brycej @LittleRabbtPaws @Twitch @awscloud @WatchMixer @Azure @ianhamilton_ 17/ If you still want to charge for it, that would still be a huge step forward from the current state of requiring broadcasters to figure it all out on their own.
/rant
/rant
