What version of the Codex App are you using (From “About Codex” dialog)?
GPT‑5.6 Sol.
What subscription do you have?
ChatGPT Plus subscription
What platform is your computer?
Windows 10 (Google Chrome browser).
What issue are you seeing?
I cannot transcribe uploaded Arabic MP3/MP4 audio files in Codex. I need Codex to extract the artist name, album title, and production company in spoken order, flag uncertain words, and write the results into an existing Excel workbook without changing its formatting.
What steps can reproduce the bug?
Open ChatGPT Work/Codex in Google Chrome on Windows 10 using a ChatGPT Plus account.
Select GPT-5.6 Sol.
Upload a long-form Arabic MP3 or MP4 audio file.
Ask Codex to transcribe the audio in spoken order and extract the artist name, album title, and production company.
Ask it to write the extracted data into an existing Excel workbook without changing the original formatting.
Codex receives the file but cannot access or transcribe the audio, so the task cannot be completed.
The same issue occurs after converting MP4 to MP3 and splitting the recording into shorter files.
What is the expected behavior?
Codex should transcribe the uploaded Arabic MP3/MP4 file accurately, preserve the spoken order, extract each artist name, album title, and production company into separate columns, flag uncertain words for review, and write the results into the existing Excel workbook without changing its original formatting, formulas, filters, column widths, or styles.
Additional information
This is a recurring professional workflow for archiving a large Arabic radio library. The test recording was approximately 15 minutes long, with one speaker clearly reading structured information. I tried MP4, MP3, and three shorter MP3 segments, but Codex still could not transcribe them. Once I provided a third-party transcript, Codex successfully corrected the text and created the Excel workbook, confirming that native audio transcription is the missing capability. Direct transcription within Codex would also avoid uploading potentially confidential media archives to third-party services.
What version of the Codex App are you using (From “About Codex” dialog)?
GPT‑5.6 Sol.
What subscription do you have?
ChatGPT Plus subscription
What platform is your computer?
Windows 10 (Google Chrome browser).
What issue are you seeing?
I cannot transcribe uploaded Arabic MP3/MP4 audio files in Codex. I need Codex to extract the artist name, album title, and production company in spoken order, flag uncertain words, and write the results into an existing Excel workbook without changing its formatting.
What steps can reproduce the bug?
Open ChatGPT Work/Codex in Google Chrome on Windows 10 using a ChatGPT Plus account.
Select GPT-5.6 Sol.
Upload a long-form Arabic MP3 or MP4 audio file.
Ask Codex to transcribe the audio in spoken order and extract the artist name, album title, and production company.
Ask it to write the extracted data into an existing Excel workbook without changing the original formatting.
Codex receives the file but cannot access or transcribe the audio, so the task cannot be completed.
The same issue occurs after converting MP4 to MP3 and splitting the recording into shorter files.
What is the expected behavior?
Codex should transcribe the uploaded Arabic MP3/MP4 file accurately, preserve the spoken order, extract each artist name, album title, and production company into separate columns, flag uncertain words for review, and write the results into the existing Excel workbook without changing its original formatting, formulas, filters, column widths, or styles.
Additional information
This is a recurring professional workflow for archiving a large Arabic radio library. The test recording was approximately 15 minutes long, with one speaker clearly reading structured information. I tried MP4, MP3, and three shorter MP3 segments, but Codex still could not transcribe them. Once I provided a third-party transcript, Codex successfully corrected the text and created the Excel workbook, confirming that native audio transcription is the missing capability. Direct transcription within Codex would also avoid uploading potentially confidential media archives to third-party services.