Single quotes in Russian

Common Voice
#1

Some Russian sentences have only one french quote, opening («) or closing (»). What I should do with them?

Examples:

  • «Посмотрим, посмотрим, – сказал он (Should be: «Посмотрим, посмотрим», – сказал он)
  • Отцы родные, отведите меня к Ивану Кузмичу». (Should be: «Отцы родные, отведите меня к Ивану Кузмичу» or without quote)

It’s punctuation problem and usually we ignore (example, missed commas). But in this case I think, that incorrect quotes is problem, when people want make audio clips, because comma and full stop are small punctuation symbols, but french quotes are big.

#2

They are most probably preprocessed by a sentence segmenter, which is not aware of them. I also coded one and it is also not aware. Therefore I manually postprocess them to correct such occurrences. But I can miss some…

Usually, during training, all these punctuations got stripped out using a normalization process. But I prefer to correct them as they can change the sound of the voice (accents, silence etc), although we tend to accept wrong intonation.

During the Sentence Collector phase, I ask the validators to reject them, so that I can correct and re-enter them to the text-corpus to prevent further problems.

If I see such mishaps during recording/listening, if it does not change the sound/reading, I disregard these errors. But that’s me…

1 Like
#3

OK, thank you. If I correctly understood, you advise to skip them like commas or full stops. Then I will do that

#4

I merely shared how I / we process them, with some insight - and it is context dependent.

The final decision is yours (+ Russian community).

1 Like
#5

Thank you again, I understood :slight_smile: