Experiments With Transcription & Translation Of Russian Television News

Last week, RT Editor-In-Chief Margarita Simonyan discussed the possibility of Russia's invasion of Ukraine leading to nuclear war in a segment on RUSSIA1 at 9:48PM Moscow Time April 26, 2022. How might a non-Russian speaker be able to find and read this statement for themselves in that broadcast?

Google's Speech to Text API supports Russian language transcription using the "latest_long" model. However, since it does not currently support native video ingest, we first have to extract the audio from the source MP4 video file. Given that these are stereo files, we follow the API's recommendations for multichannel stereo audio extraction and extract to mono right/left channels:

time ffmpeg -i ./VIDEO.mp4 -filter_complex "[0:a]channelsplit=channel_layout=stereo[left][right]" -map "[left]" VIDEO.left.flac -map "[right]" VIDEO.right.flac

Then we submit one of the channels (in this case we picked the left channel) to the API for transcription:

curl -s -H "Content-Type: application/json" \
-H "Authorization: Bearer "$(gcloud auth application-default print-access-token) \
https://speech.googleapis.com/v1/speech:longrunningrecognize \
--data "{
'config': {
'encoding': 'FLAC',
'languageCode': 'ru-RU',
'enableWordTimeOffsets': true,
'enableWordConfidence': true,
'enableAutomaticPunctuation': true,
'maxAlternatives': 30,
'model': 'latest_long'
},
'audio': {
'uri':'gs://yourbucket/VIDEO.left.flac'
},
'output_config': {
'gcs_uri':'gs://yourbucket/VIDEO.stt.json'
}
}"

This will return the OPERATIONID of the submission, we can periodically check to see when it has completed. In practice this takes just a few minutes:

curl -H "Authorization: Bearer "$(gcloud auth application-default print-access-token) -H "Content-Type: application/json; charset=utf-8" "https://speech.googleapis.com/v1/operations/YOUROPERATIONID" | head -50

Then we download the transcript from GCS to our local computer:

gsutil cp gs://yourbucket/VIDEO.stt.json .

The final transcript includes per-word confidence scores and start/stop timestamps and automatically batches the text into effective paragraphs and inserts punctuation based on language models.

This yields the following Russian transcription of that particular exchange:

К войне идем.

Не хотелось бы не хотелось бы знаешь, я где-то недавно тоже участвовала в одной такой дискуссии и один из экспертов говорил, что ну, в общем, два пути или мы проигрываем на Украине мы значит, Россия и всё. Вот это вот сдаём или начинает с третьим мировая и я его спрашиваю говорю, а вы сами какой путь считаете наиболее реалистичным, Лично я считаю наиболее более реалистичным путь третьей мировой. Ну потому что зная нас, зная нашим руководителями Путина Владимира Владимировича Зная как вообще здесь всё устроено, это невозможно. Ну то есть, ну нет, вообще шансов, что мы просто сложим лапки скажем. Вы знаете не получилось. Мы думали, что сейчас вот мы защитим, Донбасс наведём порядок, Где милитаризируем где идентифицируем. Ну что-то пошло не так не получилось, Поэтому извините мы обратно вот это вы просто не может быть.

Вот самое невероятное, что, в конце концов всё это закончится ядерным ударом мне представляется всё же более вероятным, чем вот такое развитие событий к ужасу моему с одной стороны с другой стороны к пониманию. Что ну, что, значит, так знаешь, вот это вот история с покушением произвела на меня гораздо меньше впечатление, Чем я могла бы подумать. Я, во-первых, Пользуясь случаем, Что называется какой Якубовича хочу сказать огромное спасибо ФСБ и коллегам ребятам бойцам, которые всё это вскрыли арестовали. Надеюсь что больше их там нигде не осталось, Потому что знаешь, как чай поёт не спешит и нас хоронить. У нас ещё здесь дела. У нас дома детей мало, да и просто хотелось пожить над с одной стороны.

А с другой стороны, но мы же все умрём когда-то кто-то умрёт от долгой, неизлечимой болезни кто-то умрёт от неизбежной неминуемой старости с этой точки зрения быть, застреленный за право говорить правду и защищать свою родину. Так, как ты считаешь нужным это ничуть не страшнее, чем другие варианты смерти. Вот, если был вариант, что смерти вообще нет, тогда, да, А так-то нет, поэтому я к этому отношусь так, как религии моей мне подписано, Что теперь делать с этими всеми, что должно и будет. Что делать с этими людьми, конечно, удивительно. Я обратила внимание. На что в этой всей ситуации, Что они-то Все местные все шестеро, которые вот на тебя планировали покушение на нас ссорились тигра на всей семьёй хотели. Я так понимаю уничтожить, потому что я эти гранули в списке.

Google Translate translates it as:

We are going to war.

I don’t want to, I don’t want to know, I also recently participated in one such discussion and one of the experts said that, well, in general, there are two ways or we lose in Ukraine, we mean Russia and that’s it. Here we are handing over or starting with the third world war and I ask him and I say, and you yourself which path do you consider the most realistic. Personally, I consider the path of the third world to be the most realistic. Well, because knowing us, knowing our leaders Putin Vladimir Vladimirovich Knowing how everything is arranged here in general, this is impossible. Well, that is, well, no, in general, the chances are that we will simply put our paws together, let's say. You know it didn't work. We thought that now we will protect, we will put things in order in Donbass, where we will militarize where we will identify. Well, something went wrong, it didn't work out, So sorry we're back here it just can't be.

The most incredible thing is that, in the end, all this will end with a nuclear strike, it still seems to me more likely than such a development of events, to my horror, on the one hand, on the other hand, to understanding. Well, what, so you know, this story of the assassination made a much less impression on me than I might have thought. I, firstly, Taking this opportunity, What is called what kind of Yakubovich, I want to say a huge thank you to the FSB and fellow guys, the fighters who opened all this and arrested. I hope that there are no more of them left, Because you know how tea sings in no hurry to bury us. We still have business here. We have few children at home, and we just wanted to live above on the one hand.

On the other hand, but we will all die someday, someone will die of a long, incurable disease, someone will die of inevitable inevitable old age, from this point of view, to be shot for the right to speak the truth and defend their homeland. So, as you see fit, it is no worse than other options for death. Now, if there was an option that there was no death at all, then, yes, But it’s not, so I treat this the way my religion signed me, What should I do with all of these, what should and will be. What to do with these people is, of course, amazing. I paid attention. Why in this whole situation, What are they? All six of the locals, who were planning an assassination attempt on you, quarreled with the tiger for the whole family. As I understand it, destroy, because I have these granules on the list.

While Bing Translate translates it as:

We are going to war.

I don't want to know, I recently also participated in one such discussion and one of the experts said that, well, in general, two ways or we lose in Ukraine, we mean Russia and that's it. This is what we are giving up or starting with the third world war and I ask him what path you yourself consider the most realistic, Personally, I think the path of the third world war is the most realistic. Well, because knowing us, knowing Vladimir Vladimirovich Putin as our leaders, knowing how everything works here in general, this is impossible. I mean, well, no, there's no chance that we'll just fold our paws. You know it didn't work out. We thought that now we will protect Donbass, restore order, Where we militarize where we identify. Well something went wrong didn't work out, so sorry we go back here you just can't be.

The most incredible thing is that, in the end, all this will end in a nuclear strike seems to me still more likely than this development of events to the horror of my on the other hand to understanding. Well, that means, you know, this is the story of the assassination that made a lot less of an impression on me than I might have thought. First of all, I would like to say a huge thank you to the FSB and fellow fighters who uncovered all this and arrested all this. I hope that there are no more of them left there, because you know how tea sings is not in a hurry and we are buried. We still have business here. We don't have many children at home, and I just wanted to live on one side.

And on the other hand, but we will all die someday someone will die of a long, incurable disease someone will die of inevitable old age from this point of view to be, shot for the right to tell the truth and defend their homeland. As you see fit, it's no worse than other death options. Now, if there was an option that there was no death at all, then, yes, And so no, so I treat it as my religion is signed to me, What to do with all these all that should and will be. What to do with these people is, of course, amazing. I noticed. To what in this whole situation, That they are All the local all six who planned an assassination attempt on you were arguing tigers on the whole family wanted. I understand to destroy, because I have these granules on the list.

Thus, starting with a Russian-language television news broadcast, we can completely autonomously construct a Russian-language transcription of it and translate to English!