Embed audio in Power BI report from OneDrive or Sharepoint

OwenAuger Thank you for your suggestions. The sharepoint method does not work for me since audio is confidential so the base64 string is the only way. Could you guide us how to generate the HTML2 in your pbix file please? Appreciate it.

Hi PatChan 

Sure thing 🙂

1. MP3s where entire Base64 & HTML string can be stored as a text value in a table

The method I used in the original example only works if the entire HTML string including the Base64 encoded MP3 can be stored as a text value in a column. This means the entire string must be no more than 32,766 characters (see here).

Assuming this is the case, you can use this method.

The column I originally created contains text following the below this structure where <BASE64_ENCODED_MP3> is replaced with the Base64 encoding of the binary contents of an MP3 file:

<audio controls><source src='data:audio/mp3;base64,<BASE64_ENCODED_MP3>' type = 'audio/mp3'> </audio>

If you have already loaded the binary content from source, here is a function that would create this HTML text is:

( MP3_Binary as binary) =>
let
    Base64 = Binary.ToText(MP3_Binary,BinaryEncoding.Base64),
    HTML = "<audio controls><source src='data:audio/mp3;base64,"& Base64 & "' type = 'audio/mp3'> </audio>"
in
    HTML

2. MP3s where the Base64 & HTML string is too large to be stored as a text value in a table

For longer MP3s, you will need to split the Base64-encoded MP3 data across multiple rows, similarly to how images were split in this post.

Again assuming that you have already loaded the binary content from source, you could use this function to create a table per MP3 containing the Base64 string split across rows:

(MP3_Binary as binary) =>
let
    SplitSize = 30000,
    Base64 = Binary.ToText(MP3_Binary,BinaryEncoding.Base64),
    Base64Split = Splitter.SplitTextByRepeatedLengths(SplitSize)(Base64),
    Positions = List.Positions(Base64Split),
    PartRecords = List.Transform(Positions, each [MP3 Part Index = _+1, MP3 Part Base64  = Base64Split{_}]),
    PartTable = Table.FromRecords(PartRecords, type table[MP3 Part Index = Int64.Type, MP3 Part Base64 = text])
in
    PartTable

Applying this method, you end up with multiple rows per MP3, which must be reassembled by a measure.

In my sample PBIX attached, the measure I created is:

MP3 HTML = 
VAR MP3_Index = SELECTEDVALUE ( MP3[MP3 Index] )
VAR Result =
    IF (
        NOT ISBLANK ( MP3_Index ), -- Require single selection
        VAR MP3_Name =
            SELECTEDVALUE ( MP3[MP3 Name] )
        VAR Base64Concatenated =
            CONCATENATEX (
                MP3,
                MP3[MP3 Part Base64],
                "",
                MP3[MP3 Part Index]
            )
        VAR MP3HTML =
            "<label for='audio-player'>" & MP3_Name & "</label>"
            & "<audio id='audio-player' controls><source src='data:audio/mp3;base64,"
            & Base64Concatenated
            & "' type = 'audio/mp3'> </audio>"
        RETURN
            MP3HTML
    )
RETURN
    Result

As an example, I used the HTML visual to display the HTML per MP3:

Hope that helps!

Regards,

Owen

Thank you OwenAuger for responding so quickly! I managed to generate the binary64 string for my 4-sec recording using the python code below. It's surprising that a 4-sec recording can be translated to such a long string. Anyway, the recording got cut off sadly after 1-2 seconds. I am not sure why 😕 Please don't feel obliged to respond as I think this should be a separate thread already!