Forvo.com sound data base or the ultimate language learning plug-in for Anki (project idea)
mobiled...@googlemail.com
I just love this Forvo Website for hearing native speakers pronounce all the words you want. It is so simple as if there is any word in any language of this world missing you just add it and a few hours later some native speaker already recorded it for you and the rest of the community. http://www.forvo.com
If you could merge those features of Anki audio integration with the forvo.com audio files it would just be awesome, if the add-on does not find the audio by a native speaker at forvo.com to simply add it so someone will record it immediately on forvo.com. I guess forvo.com becomes in that way the most powerful database as well. This automatic process would create an awesome functionality to the SRS of Anki with full native speaker pronunciation of all the words in the world.
Let me know if you got into this project idea. I would be totally thankful for that forever and with me of course all those people in the world sucking up languages pronunciation. This way you can review your vocabulary while listening to native speakers and repeating after them in order to train your pronunciation! I am excited.
Keep my updated.
Andreas
Nicolas Raoul
Forvo sound files are Creative Commons BY-NC-SA, so we can use them indeed.
Could be an alternative to TTS (voice synthesis) for single words.
Their API is free for 1000 calls per day.
This is too low, we have two options:
- Maintain a mirror (might be tricky with 1000 calls per day)
- At first use, ask the user to enter her own API key.
A nice thing is that for common words, usually they have different
speakers, sometimes with different accents.
This could enable us to develop the voice equivalent of random fonts:
http://code.google.com/p/ankidroid/issues/detail?id=1044
Should we save the sound file to media, or download it each time?
I created this issue to keep track of that idea:
http://code.google.com/p/ankidroid/issues/detail?id=1092
Let's continue the conversation there.
Anyone interested please "start" this issue.
Cheers!
Nicolas Raoul
> --
> You received this message because you are subscribed to the Google Groups
> "AnkiDroid" group.
> To view this discussion on the web, visit
> https://groups.google.com/d/msg/anki-android/-/YRGsUC48bM0J.
> To post to this group, send an email to anki-a...@googlegroups.com.
> To unsubscribe from this group, send email to
> anki-android...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/anki-android?hl=en-GB.
joed...@excite.com
Nicolas,
Hallo aus New Jersey! Wie geht's dir?
Regarding Forvo - some background on my past experiences:
I've used Forvo since it 1st went on-line back in early 2008, for my target languages:
German, Norsk-Bokmål, Norsk-Nynorsk, Swedish, Finnish, Tagalog, Vietnamese.
I've attempted to contact Forvo on numerous occasions regarding the use of "CSV structured file formats" which could be used to batch load new word recordings by native speakers, and to help reduces the number of pending word lists.
Two yrs ago, I had a group of 10 Norwegian students that were studying here in NJ und who wanted to help with creating new recordings. After a week of problems I emailed Forvo but they never responded. The students tried a few more words ... but the interface was so bad the recordings sounded like they were done in a tin can. When we tried the recordings on a few other sites it worked great ...the play back was so clear. We tried for over a month and it just was too frustrating scheduling studio recording time and getting them together ... Once I realized that Forvo wasn't going to reply they went on to another local Son's of Norway project.
I've also written Forvo about some plug-in problems and a "Bookmarklet" problem with a button feature. No response either!
Depending on the target languages you had in mind ... maybe a project could be organized to assist in the capture of the .mp3 files.
Using German as an example, we're talking about 94,961 recordings und 324 pending.
- Assuming that all 94,961 recording are correct ... a group of 100 users (downloading 100 .mp3 files per day including time to modified the mp3 file descriptors to replace number references) could recover most of that in less than 6 months. But you would still need native speakers to rate und discard those .mp3's that are just poorly recorded or outright wrong.
It might be an interesting & very effective project if planned out and executed correctly.
If you would like to discuss this in more detail let me know.
Danke sehr!
Joe
On Thursday, April 5, 2012 9:51:42 AM UTC-4, Nicolas Raoul wrote:
Hi Andreas,
Forvo sound files are Creative Commons BY-NC-SA, so we can use them indeed.
Could be an alternative to TTS (voice synthesis) for single words.
Their API is free for 1000 calls per day.
This is too low, we have two options:
- Maintain a mirror (might be tricky with 1000 calls per day)
- At first use, ask the user to enter her own API key.
A nice thing is that for common words, usually they have different
speakers, sometimes with different accents.
This could enable us to develop the voice equivalent of random fonts:
http://code.google.com/p/ankidroid/issues/detail?id=1044
Should we save the sound file to media, or download it each time?
I created this issue to keep track of that idea:
http://code.google.com/p/ankidroid/issues/detail?id=1092
Let's continue the conversation there.
Anyone interested please "start" this issue.
Cheers!
Nicolas Raoul
Nicolas Raoul
Sorry to hear that Forvo has problems :-/
Also, their license is good but there does not seem to be any way to
download a dump of all data (correct me if wrong).
I am a huge fan of Wikipedia/Wiktionary/WikiCommons, and I am sure it
would be the best place for this content.
Actually, they already do this, even though it is not user-friendly.
See for instance:
http://en.wiktionary.org/wiki/thorough
You can see they have the pronunciation, for both US and UK.
Their database is already quite impressive, see for instance:
http://commons.wikimedia.org/wiki/Category:English_pronunciation
A great project would be to create a user-friendly web site to upload
missing pronunciations.
Joe, do you think you could start this project?
It could be a JavaScript plugin within the user's vector.js, or a
separate website, or a standalone software. Are you familiar with
those kind of tools? I can provide examples for inspiration.
You can brainstorm the project here and recruit volunteers:
http://en.wiktionary.org/wiki/Wiktionary:Beer_parlour
http://en.wiktionary.org/wiki/Wiktionary:Mailing_lists
Andreas, you seem very enthusiastic about pronunciation collaboration,
I am sure you could be even more involved with Wiktionary
pronunciations :-)
By the way:
In my last email I wrote that it could be an "alternative to TTS" but
5 minutes after I realized that using TTS to speak a foreign language
is not a good idea. TTS is acceptable if pronunciation is not the
topic of study (OK for grammar or non-language related study though).
To study pronunciation, nothing can replace real people's voice.
Cheers!
Nicolas Raoul
>> > anki-android...@googlegroups.com.
>> > For more options, visit this group at
>> > http://groups.google.com/group/anki-android?hl=en-GB.
>
> --
> You received this message because you are subscribed to the Google Groups
> "AnkiDroid" group.
> To view this discussion on the web, visit
> https://groups.google.com/d/msg/anki-android/-/YTtXD_gHCa4J.
>
> To post to this group, send an email to anki-a...@googlegroups.com.
> To unsubscribe from this group, send email to
> anki-android...@googlegroups.com.
joed...@excite.com
I tested a few things tonight und was able to easily create an Excel Spreadsheet of approx 250 Forvo words. The spreadsheet is attached for review.
Due to an excessive number Forvo Gateway Timeouts, I was only able to download 54 mp3 recording. If I factor out the Gateway Timeout issue, it took approx 6 seconds for each download. The spreadsheet with data took approx 5 minutes to create. There’s extra info on the spreadsheet that can be deleted once the downloaded MP3's are renamed.
What are your feelings on the scope & structure of translations, verb conjugations, example sentences, etc? I would think that once an agreed upon structure is established for these issues, the recording und data gathering process will be more efficient. We should also set some sort of benchmark for when or how often the translations are entered into the database. Maybe it could be done in groups of 100. If it lags to far behind ... playing catch up will be a mess.
An other idea would be to enlist the help of some native speaker's to generate a series of basic but useful "Example Sentences". Maybe even some pronunication drills and reps.
We could start off with a couple of target languages und see how smoothly it goes. I'd be willing to assist with German und Norsk. If we have enough people involved doing the downloads and MP3 re-names maybe we can figure a way to create a keyed relational database with cross referencing to other target languages. We could also use the RSS feed to watch for updates from some of the top 5 Forvo Users in each target language.
I'll continue the downloads tomorrow and refining the process as I go.
Danke sehr.
Hi Joe,
Cheers!
Nicolas Raoul
>> > For more options, visit this group at
>> > http://groups.google.com/group/anki-android?hl=en-GB.
>
> --
> You received this message because you are subscribed to the Google Groups
> "AnkiDroid" group.
> To view this discussion on the web, visit
> https://groups.google.com/d/msg/anki-android/-/YTtXD_gHCa4J.
>
> To post to this group, send an email to anki-a...@googlegroups.com.
> To unsubscribe from this group, send email to