Post ANpG9pCc4bTGnuBeim by Cambria@fosstodon.org
(DIR) More posts by Cambria@fosstodon.org
(DIR) Post #ANpCEEU7cGXQYdDsgq by Gina@fosstodon.org
2022-09-22T14:08:20Z
0 likes, 2 repeats
#fediverse could I please ask you for recommendations for automatic transcribing software?I don't care if it's closed source, costly or if they ship my data to Xi Jinping himself, all I care about is that it's ACCURATE AF. It's for my thesis so it's needs to capture every eh and uhm to perfection. 🤌
(DIR) Post #ANpDnUb1o2z0cEpXd2 by rimugu@liberdon.com
2022-09-22T14:25:57Z
0 likes, 0 repeats
@Gina I have used (I am ashamed of it) MS Word.It has a dictate feature.If what you want is already recorded you can use a virtual cable (e.g. https://vb-audio.com/Cable/) to set your recording to the virtual output and word to the virtual input to connect them.
(DIR) Post #ANpEOaaAVquntc7Kbo by kayb@chaos.social
2022-09-22T14:31:48Z
0 likes, 0 repeats
@Gina do you mean OCR?
(DIR) Post #ANpFWhwT8KfGzaz1No by joel@fosstodon.org
2022-09-22T14:45:14Z
0 likes, 0 repeats
@kayb @Gina speech to text.My recommendation?Get a pixel :blobcatderpy:
(DIR) Post #ANpFeEjrav44jjg34S by joel@fosstodon.org
2022-09-22T14:45:37Z
0 likes, 0 repeats
@kayb @Gina speech to text.My recommendation?Get a pixel 6 :blobcatderpy:
(DIR) Post #ANpG9pCc4bTGnuBeim by Cambria@fosstodon.org
2022-09-22T14:52:19Z
0 likes, 0 repeats
@Gina Some of the radio nerds at work were talking about this this morning and said it's better than the AWS offering. This was for recordings in English. https://openai.com/blog/whisper/
(DIR) Post #ANpGAScrUwdn4blfY8 by gunsrude@stellar.build
2022-09-22T14:52:27Z
0 likes, 0 repeats
@Gina Pretty sure the Dragon stuff from Nuance is still in the top few. Super pricey though.
(DIR) Post #ANpGSjPDqhwIKP6H5c by baslow@mastodon.social
2022-09-22T14:55:48Z
0 likes, 0 repeats
@Gina Otter.ai (regrettably *not* open source) has done a good job for me. It filters out "ehs" and "ums" however. Since it is mainly geared to transcribing meetings (including speaker identification) for large organizations it can afford to offer a free tier (several hours a month) to individuals. If your laptop mic is good enough you can use the web interface but there is an Android app (don't know about iOS)
(DIR) Post #ANpIVcmz6eNeaFipqC by hq1@fosstodon.org
2022-09-22T15:18:39Z
0 likes, 0 repeats
@joel @kayb @Gina totally unhelpful reply but my favorite is real time subtitles at Google meet. With people from all over the world it's so inaccurate even the most boring meeting becomes comedy gold.
(DIR) Post #ANpIyxyZPfSwgw8GwK by Gina@fosstodon.org
2022-09-22T15:23:55Z
0 likes, 0 repeats
@hq1 @joel @kayb haha love that, thanks, I'm looking for transcribing software though, not translating.
(DIR) Post #ANpVHGpBGvnpoOEdYO by fcovicente@fosstodon.org
2022-09-22T17:41:36Z
0 likes, 0 repeats
@Gina Hi! Open AI (frequently very closed) just open soursed their state-of-the-art speach to text model: https://github.com/openai/whisperI haven't tried it yet but I heard it's very good.
(DIR) Post #ANpVbVuCuWWSNPEs7s by hq1@fosstodon.org
2022-09-22T17:45:29Z
0 likes, 0 repeats
@Gina @joel @kayb I continue to be unhelpful but those subtitles are transcriptions. By "people worldwide" I meant there's variety of accents that make it go nuts
(DIR) Post #ANpdIsuZwxq5rM2aIa by reil@mastodon.social
2022-09-22T19:11:44Z
0 likes, 0 repeats
@Gina I can't speak to its quality, but I know that Descript is used by a decent number of people. Might as well throw that into your list to look at!
(DIR) Post #ANpeYHGnlgVFhhL54C by sarvasana@fosstodon.org
2022-09-22T19:25:42Z
0 likes, 0 repeats
@Gina Android has a Live Transcribe function.
(DIR) Post #ANpihmbN2w51IgkR7o by kai@ajin.la
2022-09-22T20:12:13Z
0 likes, 0 repeats
@Gina Google understands me even if I mumble while brushing my teeth
(DIR) Post #ANppUKXgl5qiBrYfMe by K4mpfie@mastodon.social
2022-09-22T21:28:14Z
0 likes, 0 repeats
@Gina Well if you want to catch all the um's and hmm's you probably be best advised to go to a human transcriber. Another fairly good solution is Microsoft Word 365 browser version that can even differentiate between speakers and does roughly 1h of Interview in 10 minutes
(DIR) Post #ANpuDnvujNNUBxpQiu by gerald_leppert@bonn.social
2022-09-22T22:20:44Z
0 likes, 0 repeats
@Gina @Gina Very good for automatic interview #transcription in English and German language is #f4x https://www.audiotranskription.de/f4xand it is #GDPR compliant.
(DIR) Post #ANq80exECXhBwF4Mee by mrkz@mstdn.mx
2022-09-23T00:55:47Z
0 likes, 0 repeats
@Gina there is https://us.ankerwork.com/pages/transcription I haven't used it, but I considered it a while ago and there are some reviews to check on YouTube about it
(DIR) Post #ANqUY57s7ABJekOEtM by nosat@liberdon.com
2022-09-23T05:08:22Z
0 likes, 0 repeats
@Gina if you use linux then festival is OK, it integrated with LibreOfficehttp://festvox.org/festival/
(DIR) Post #ANqgEG4mtIulHBGG80 by me@social.jlamothe.net
2022-09-23T01:52:59Z
0 likes, 0 repeats
Honestly, if accuracy is the goal, you're gonna want a human transcriber. If Google can't even get it right (see auto-generated YouTube captions) I doubt if anyone can.I used to do subtitling for Rev.com. They do decent work, though their style guide directs to omit things like false starts and such. They also do audio transcription.The catch is that they classify their workers as "independent contractors" so they don't have to pay a living wage. So there is that ethical dilemma.
(DIR) Post #ANqgEGlKL8wTP7AEQy by Gina@fosstodon.org
2022-09-23T07:19:04Z
0 likes, 0 repeats
@me that and that it's expensive 🥲
(DIR) Post #AOtz6ksbWT9bdd5EVU by rogbeer@im-in.space
2022-10-24T19:26:59Z
0 likes, 0 repeats
@Gina I know my reply comes a little late.But someone has told me about https://www.perapera.ai/translation-services/Not too sure how much it costs