Can anybody decipher this pilot's landing request due to moderate turbulence (some unreadable words...). Thanks ! https://www.icloud.com/iclouddrive/03kDM9-xQ7Oyz3kUaeRJj6M5Q#AUDIO-2021-12-03-07-51-16


2 Answers 2


This is what I hear.

“Boblika tower, Air China Niner Two Niner request”

“Air China Niner Two Niner request to uh climb to eleven thousand six hundred meters due to uhh moderate turbulence”

It is the same person making both transmissions as you don’t hear the tower respond to the first request.

The tower name is most probably incorrect, but if you went through a list of control towers you might be able to pick it out and correct the spelling.

It may not even be a tower that the request is made to. It seems unlikely a tower facility would be controlling flights operating at 11,600 meters, but in Russia or China, anything may be possible.

  • $\begingroup$ Sweet. Pretty much a copy of my answer 😃 $\endgroup$
    – Jpe61
    Dec 5, 2021 at 15:18
  • $\begingroup$ Thank you very much for helping on this one ! I was understanding a request to land so I was confused with the altitude mentioned, $\endgroup$
    – Brani
    Dec 5, 2021 at 20:24
  • 1
    $\begingroup$ Thank you very much to both of You, Mike and Jpe, for helping on this one ! I was understanding a request to land so I was confused with the altitude mentioned. Now, I "clearly" understood that it was a request to climb. Thanks again ! I'm not myself a native english speaker but was pretty baffled by the pilot's accent ! As for the facility called by the crew, I looked at the planned route of Air China 929 ( Shangai PVG to Tokyo NRT) and the associated facilities: I found that KCZ Kōchi airport might be the one called, but the first word of the pilot comm is totally unreadable. $\endgroup$
    – Brani
    Dec 5, 2021 at 20:41
  • $\begingroup$ Anyway, Thank you very much again, as I now understand this communication ! Cheers ! $\endgroup$
    – Brani
    Dec 5, 2021 at 20:41

If anybody's interested, here is what OpenAI's Whisper transcription model produces, verbatim:

Air China 939 request to climb to 11,600 meters due to moderate turbulence.

Looks like it filters out the filler noises and pretty accurate.

Here is the Python script:

import sys
import openai

openai.api_key = # YOUR_OPENAI_API_KEY

filename = sys.argv[1]

audio_file = open(filename, "rb")

result = openai.Audio.transcribe(
  prompt="Transcribe this airplane communication audio file, differentiate between speakers where possible. Return markdown format."

# Save to Markdown file
out_file = filename.replace('.mp3', '.md')
with open(out_file, 'w') as f:

print(f"Transcription saved to {out_file}")

Note: I'm not sure the prompt does anything, I tried to get a nice markdown format out of it, but it just produces the plain text. Maybe too short, but would need to try with more files.

  • 1
    $\begingroup$ I like this answet $\endgroup$
    – U_flow
    Jul 23, 2023 at 15:32

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .