Assistance Required: Acoustic Data Import Failure

Question

Dear Azure Support Team,

I hope this email finds you well. I am reaching out regarding an issue with uploading and processing acoustic data in the Speech Service. Despite following all recommended steps, the import fails with the error message:

"Acoustic data import failed: Zero transcriptions could be parsed from the given input."

Below is a detailed summary of what we have tried so far:

1. File Standardization

• We ensured that all .wav files are in mono format with a sample rate and bit depth matching Azure’s requirements.
• We converted the audio using both SoX and FFmpeg to avoid formatting discrepancies.
• Audio lengths were verified to ensure no file exceeds Azure's limits.

2. CSV File Verification

• We checked that the CSV (transcriptions.csv) contains two columns: file_name and text.
• The file encoding is UTF-8 with BOM as per the documentation.
• We removed duplicate rows, missing values, and ensured each audio file in the folder has a matching transcription entry in the CSV.

3. Chunking Large Folders

• We split the large dataset into smaller chunks with a size of around 8MB per zip file.
• Each chunk contains its corresponding audio files and a correctly formatted transcriptions.csv.

4. Scripts Used

We wrote and ran Python scripts to:

• Validate file formats, CSV headers, and encoding.
• Reformat and fix CSV issues.
• Verify that all .wav files match the entries in the CSV.
• Split the dataset into smaller, manageable chunks.

5. Azure Upload Attempts

Despite these efforts, each upload still results in the same error mentioned above, preventing us from moving forward with the project.

---

We kindly request your guidance to help identify the issue. Are there additional requirements or validations that we might have missed? I am happy to provide sample data, including audio files and the CSV, if needed for further troubleshooting.

Thank you for your time and support. I look forward to your response.Dear Azure Support Team,

I hope this email finds you well. I am reaching out regarding an issue with uploading and processing acoustic data in the Speech Service. Despite following all recommended steps, the import fails with the error message:

"Acoustic data import failed: Zero transcriptions could be parsed from the given input."

Below is a detailed summary of what we have tried so far:

1. File Standardization

• We ensured that all .wav files are in mono format with a sample rate and bit depth matching Azure’s requirements.
• We converted the audio using both SoX and FFmpeg to avoid formatting discrepancies.
• Audio lengths were verified to ensure no file exceeds Azure's limits.

2. CSV File Verification

• We checked that the CSV (transcriptions.csv) contains two columns: file_name and text.
• The file encoding is UTF-8 with BOM as per the documentation.
• We removed duplicate rows, missing values, and ensured each audio file in the folder has a matching transcription entry in the CSV.

3. Chunking Large Folders

• We split the large dataset into smaller chunks with a size of around 8MB per zip file.
• Each chunk contains its corresponding audio files and a correctly formatted transcriptions.csv.

4. Scripts Used

We wrote and ran Python scripts to:

• Validate file formats, CSV headers, and encoding.
• Reformat and fix CSV issues.
• Verify that all .wav files match the entries in the CSV.
• Split the dataset into smaller, manageable chunks.

5. Azure Upload Attempts

Despite these efforts, each upload still results in the same error mentioned above, preventing us from moving forward with the project.

---

We kindly request your guidance to help identify the issue. Are there additional requirements or validations that we might have missed? I am happy to provide sample data, including audio files and the CSV, if needed for further troubleshooting.

Thank you for your time and support. I look forward to your response.

Answer 1

Hi Zee

Welcome to Microsoft Q&A Forum, thank you for posting your query here!

I understand that you are encountering an issue, try testing with a smaller sample size by uploading only a few audio files and their corresponding transcriptions. Make sure there are no files with zero length or silent audio as this might cause the parsing to fail.

If the issue still persists, please create a support ticket.

Thank You.