Any chance to get it implemented? I wonder if enough users care maybe we can get the creators to reconsider the decision? Or at least give hope that it’ll be done in not too distant future? It doesn’t look too complicated code wise, just create a new gender, Female voice 1, or Original female voice, whatever. For me personally, the new voice is too AI like, and Haru’s voice was one of the important factors I became a user. I even started N2 grammar recently (way too early) only to have the voice stripped underway. Anybody feel the same way?
18 Likes
誠に!
5 Likes
Due to the generally low audio quality of the original recordings (background noise etc), we feel that these recordings did not reflect the level of professionalism that we would like to keep across the platform as a whole. We can potentially make the voice packs with Haruna’s voice available for download though (I have done this previously just through my Google Drive).
For the vocab audio that we will be recording soon, we will also be asking the voice actress to mix up the tone a bit and insert more natural pauses etc so that it feels a bit more life like. The current grammar recordings, while they are very good and have a lot of fun character voices as well, there are quite a few that feel very much like something that you would hear on the JLPT, which in and of itself is not a bad thing.
13 Likes
I can definitely hear the better quality of the new recordings, but the original female audio is good enough to be available too, even if there is a noticable contrast. You could name it “legacy” or something. Definitely not “not professional enough”, I would say.
You’re right about the original male voice though. They deserve to be replaced. 
9 Likes
Honestly speaking, I don’t see where the new audio is actually better in terms of audio quality. Was the old audio…low bitrate? Had background noise? I can recall hearing a few (I think from the male voice) where it sounded like the guy was sitting too far from the mic. But it wasn’t much.
Something else…I may be in the minority here, but…aren’t the new voices, rather, anime-y? I study Japanese for professional purposes, and I prefer to hear an actual natural tone. Now, I totally appreciate that many people maybe prefer the more anime-style voice (apologies if this isn’t the best way to put it), but I can’t say I prefer it this way tbh.
6 Likes
Would be great, especially if they organized according to grammar points they belong to and are easy to locate! Also, newer Haru’s recordings seemed to be of crisp quality, probably due to better mic. But alas, they’re gone as well 
2 Likes
Haru Fan Club! I know we’re already spoilt with the daily diary with voice recordings (I still have a LOT to catch up), but I think some of us prefer the “natural” vs. “professional” in some, if not all, of the lessons. Not throwing shade at the new VA either, as I think her voice is very nice!
What do we need to do to make it happen? Set a GoFundMe to pay Haruna for extra time? Promote BunPro to new Japanese learners? I don’t have any good ideas at the moment, lol.
8 Likes
I feel the same way. No disrespect to the new voice actress (I admire the high quality voice acting), but I feel that they are sometimes too theatrical and even anime-like for my purposes, which is professional Japanese. I prefer having the more “natural” sound.
5 Likes
Did you actually A / B test the audio recordings or run any other subjective evaluation tests with some user groups (new users, existing users) before you made the decision to remove all existing audio?
And with that I just don’t mean to test the recording quality, but the overall impression of the audio. I agree that the recording quality of the new audio is much better, but for other audio qualities, I share the sentiment of the other users here in this thread that Haru’s voice is more natural.
4 Likes
We did quite extensive testing within the team of both native Japanese speakers and native English speakers on what would be the most appropriate for learners. There is quite a mix amongst all of the sentences, some have a more anime feel, some have a more professional feel, some have a newscaster kinda feel, they are mostly adjusted based on what the voice actress felt the overall tone of the sentence was.
We feel that the current style is a good mix of what would be heard on the JLPT (clear and purposefully enunciated speech), and what would be heard in real life, also with quite a few fun ones thrown in there with character voices.
Edit - It might also be worth noting that the new VA is actually joining the Bunpro team on a more ongoing basis. So we can always rerecord any specific sentences that feel a bit too professional.
6 Likes
I myself am waiting anxiously for the VA rework of vocab sentences
4 Likes
I honestly think the auto generated voice is pretty good, but some Kanji are very often misreading (君、町). At least it’s a natural tone instead of an anime-style voice.
3 Likes
Yeah I agree that the TTS is good, especially if you consider it is auto-generated. But I don’t mind the interpretation to make it more interesting.
1 Like
Well, I’d say TTS is ok-ish for short sentences, but the speech / elocution / diction becomes really awkward with longer ones such as this one, for 過ごす :
" 今日も元気に過ごしている子供たちがいつか事故や病気で死んでしまうかもしれないと想像するだに胸が苦しくなります。"
3 Likes
I don’t feel like Haru’s recordings were any worse than the current male recordings quality wise. Throwing them away seems counter productive and reduces value of this product. If the team hates them so much, they can be hidden behind an option in settings or something.
3 Likes
I’ve been trying to give the new recordings a chance over the past month or so, but honestly at this point I just don’t even play the audio. Haru’s recordings were one of the main points that attracted me to Bunpro when I first tried it. In my opinion, the value of recordings like this to the user in a language learning service, and by extension the professionalism that the recordings contribute to the service, comes from two criteria:
1. Practical use
Haru’s recordings sound like a real person (as in not a fictional character) just saying the sentences while sitting in a normal room. The new recordings certainly have a more professional acting feel, and the audio has the distinctive sound of someone speaking very close to an expensive microphone in a studio with perfect mixing and mastering. But this is not how Japanese sounds when having a real life conversation. Other Japanese learning resources often have overly professional, unrealistic recordings, and I always thought Haru’s recordings were much more practical and set Bunpro apart, so I hate to see Bunpro going in that direction too.
2. Pleasantness
From the standpoint of pure audio quality, the new recordings are “better,” but I think it’s too much. Some people have said they’re too loud, but turning down my volume, I think the issue goes beyond that to compression and EQ and such. Maybe this kind of audio would be appropriate for some sort of narration in anime over background sounds, or if it was the lead vocal in a song, but to listen to completely on its own for language learning purposes, it sounds too squashed and powerful (beyond just the volume) to be pleasant to listen to for me. Maybe even more than the audio itself, I also just really like the sound of Haru’s voice, which of course is very subjective, but it sounds like many other users feel the same. I never thought her recordings were lacking in quality in any way. They seemed fine and clear, and if anything, more pleasant to listen to because they sound closer to hearing someone’s voice face to face as opposed to very edited audio.
This is not to say anything bad about the new voice actor or audio engineer—everyone is clearly very good at what they do. And of course, I’m grateful to the Bunpro team for continuing to put in work on Bunpro. This is just not the place for that kind of audio and acting, at least for me as someone learning Japanese for the purposes of professional and everyday use. Other users may feel the opposite. But please at least just give us the option to switch back. I really don’t think that would be less professional in any way. If anything, I think it would be more professional, considering the number of users that seem to want this.
4 Likes
As if my mind was being read. Hopefully our voices will be heard and won’t remain voices in the desert. Having more available voices only adds more value, so it’s a win/win for everyone.
1 Like
Just adding that I prefer the old one too, and would very much appreciate an option!
2 Likes
賛成だよ!
1 Like
I also prefer the original voice recordings as well. It would be nice to have an option to go back.
2 Likes