Podcast episode over the upload limit
The school's digital media class publishes a 20-minute podcast episode on the school website each term. This term's episode includes a phone interview with a local MP, a music bed, and a segment recorded in the gym during a basketball game. The current export is a 48 kHz, 16-bit stereo WAV, about 230 MB. The website rejects uploads over 100 MB. Listeners report that voices are quiet compared with the music, and that the gym segment is painful to listen to. The MP's office has asked for a transcript, and the school's accessibility guidelines require one. The music bed is a Creative Commons track that requires attribution. The class has one 50-minute lesson to fix the episode and publish it, using the same laptop and free software.
1. Explain how to get the file under 100 MB without making the voices hard to understand. Include the settings you would change and why.
Read the worked answer
File size is bitrate × time, so the way to shrink a 20-minute episode is to reduce the bitrate and the number of channels, not to cut the content. A 128 kbps mono MP3 for 20 minutes is about 19 MB, well under the 100 MB limit, and speech at that rate is clear. Export as MP3 or AAC, mono, 44.1 kHz, 96–128 kbps; for speech, mono is the biggest saving because it halves the data and listeners do not lose anything useful.
Do not keep 48 kHz stereo just because the source was a WAV. Higher sample rates and stereo do not improve a single voice, and the WAV is uncompressed, which is why it is 230 MB. Check the exported file by listening on phone speakers and headphones, and confirm the upload actually succeeds rather than assuming it will. Keep the original WAV as the archive master so the episode can be re-exported later at a higher rate if needed.
2. The voices are quiet compared with the music and the gym segment is harsh. Describe the processing order you would use and one mistake to avoid.
Read the worked answer
Work on the problems in order, because each step changes what the next step hears. First, repair the gym segment: apply noise reduction gently, a high-pass filter around 80–100 Hz, and light compression, and check for artefacts. Second, balance the voice tracks to each other so the MP and the students sit at a similar level. Third, set the music bed under the voice, ducking it to roughly -18 to -20 dB below speech, with fades at the start and end. Finally, put a limiter on the master with peaks around -1 dBFS, and aim for a podcast loudness near -16 LUFS.
The mistake to avoid is normalising or compressing the whole mix first. That raises the music along with the voice, so the balance problem remains and the limiter works harder. Fix the sources and balance first, then master.
3. The MP's office wants a transcript and the track needs attribution. Explain how the class should meet both requirements within the lesson, and why each matters.
Read the worked answer
Both tasks can run in parallel. One pair runs the episode through a free automatic transcription tool, then checks it against the audio, correcting names, party names, numbers and any quote that could be misread. The transcript goes on the episode page as text, not only as a PDF, so screen readers and search engines can use it. Another pair writes the attribution line: track title, creator, licence name and version, a link to the licence, and a note that the track was altered if it was cut or faded.
The transcript matters because the accessibility guidelines require it and because a misquoted MP is a real reputational risk; automatic tools regularly mishear names. The attribution matters because the Creative Commons licence is conditional - using the track without credit breaks the licence. Build the credit into the episode description and the end of the audio, then check both before publishing, because the upload limit only matters if the episode is allowed to stay online.