The best way to edit out announcements from a radio recording
Radio recordings often contain more than the music you want to keep. Station identifications, sponsor messages, news updates, traffic reports, promotional jingles and presenter links can interrupt a performance or occupy valuable space on a CD. Some also cover the opening or closing notes of a song. Removing them requires careful judgement, especially when the transition cannot be located from silence alone.
A waveform-led manual edit gives the cleanest result. Start with a preserved copy of the original, identify each unwanted section, and work with short fades rather than abrupt cuts. This method suits recordings captured from vinyl, cassette, AM, FM or digital radio, and it works whether the final destination is a personal CD, a lossless archive or a compressed file for everyday listening.
Decide what counts as an announcement
First decide which interruptions should disappear. Most listeners want to remove advertising, station IDs, news and weather segments, traffic information and recorded promotional material. A spoken introduction by the station’s announcer may also be unnecessary if the recording begins with the performer talking.
Australian broadcasts commonly contain material from ABC Radio, SBS Radio, triple j, commercial networks and independent community stations. A public broadcaster might place a news bulletin between programmes, while a commercial break can carry sponsor reads for Coles, Woolworths, Bunnings, Kmart, AAMI or local businesses. These can sound very different, but their spoken voices share a frequency range that is usually easy to distinguish on a detailed waveform.
Set a consistent rule before editing. Some listeners keep one station identification so the source remains recognisable, while others remove every spoken item. Australian announcers often deliver sponsors briskly, sometimes with a pronounced local accent, so familiarity should not be the only guide. Listen for the complete sentence and the musical material surrounding it. This prevents a clipped word, an incomplete sponsor credit or the first line of a song from remaining accidentally.
Build a safe editing workspace
Duplicate the source recording and retain the untouched file. If the broadcast was recorded to cassette or vinyl, transfer it using the best available capture route and avoid repeated analogue-to-digital conversions. Copy the resulting WAV or FLAC file into a separate editing project, or create a clearly named duplicate if the software works directly with the original.
Audacity provides a capable free option, while REAPER, Adobe Audition and other dedicated editors offer more advanced controls. Display the waveform vertically if that makes speech easier to inspect. Keep the recording at its native sample rate and bit depth during editing. Converting a 16-bit source to 8-bit or repeatedly saving compressed copies can reduce the quality of quiet announcements and delicate cymbals, making their boundaries harder to identify.
Cassette hiss and vinyl crackle can obscure transitions, but aggressive noise reduction may damage voices or musical detail. Apply it sparingly, and preferably on a duplicate track. Leave the programme material at a consistent level while marking edits. If the recording will eventually join other tracks on a CD, the site’s guide to normalising compilation volume explains how ReplayGain can balance sources without forcing every track to the same loudness.
Mark precise edit points
Listen through the complete recording first and note each interruption as you go. Use labels or cue markers to record its approximate start and finish. A second pass at normal speed confirms the content, while a slower preview reveals whether a voice overlaps the musical introduction or outro. Do not rely entirely on automatic silence detection: announcers frequently speak over a music bed, and a quiet passage can resemble a pause between words.
Zoom into each labelled region and examine the waveform. Speech normally creates irregular clusters of peaks, whereas sustained instrumental material has a steadier shape. Jingles often have a recognisable introductory sound and an abrupt final note. Place the first marker just before the unwanted sound and the second after it. Include a little surrounding audio at both ends, as this gives you room to adjust the cut once you hear how it works.
Do not attempt to recreate music that was completely obscured by a spoken announcement. Time-stretching a covered passage usually produces an artificial result, and stretching can alter the pitch and character of the instrument. If an announcer talks over the only available introduction to a track, either retain a short section or source another performance. A precise fade around the available music will sound more convincing than an ambitious but defective reconstruction.
Remove speech and seal the joins
Split the recording at the marked boundaries and remove the unwanted section on a duplicate track. This keeps the original material available if an edit later proves inaccurate. In waveform editors, a short fade at each edge is safer than a hard cut. Even a few milliseconds can prevent the sharp discontinuity that produces a click, particularly where the waveform was previously at full level.
A crossfade of roughly 10 to 30 milliseconds is often enough to join two pieces of continuous music. Avoid a long crossfade because overlapping announcements or different musical passages become audible. Where a jingle runs into a song, shorten the jingle and fade it beneath the first clean phrase. Where the announcer stops immediately before a note, trim the silence and align the waveform rather than adding an unnecessarily broad fade.
Do not use noise reduction to disguise a poorly chosen edit. It can reduce hiss but will not remove clear speech, and repeated processing may make the join sound thin. A sustained electrical buzz is a different fault altogether. For background on the distinction between ordinary interference and spark-gap discharges, Tesla coil history explains why a continuous radio-frequency buzz should not be treated like an announcement boundary.
Check, export and keep the finished recording
Review the edited recording without skipping through the join points. Listen at normal speed through every transition, then inspect the waveform for unexpected silence, abrupt level changes and clipped peaks. Headphones reveal clicks and faint speech, while a speaker can expose edits that sound abrupt on small earbuds. Compare the result with the preserved original whenever a passage sounds suspicious.
Export a master file rather than repeatedly editing the same compressed copy. WAV or FLAC is suitable for an archive and for later CD production. For a conventional audio CD, use 44.1 kHz, 16-bit stereo PCM and check the finished disc by playing it on more than one player. Include the source date, station and track details in the file name; the day-month-year format commonly used in Australia avoids confusion with numerically formatted dates. For context on the person behind this hobby-based collection of vinyl, cassette and CD projects, Allen’s background provides useful background.
Open the source recording now, create a working duplicate and mark the first announcement before making any permanent change. Test that edit, listen to the complete result once on headphones, and then continue through the recording while keeping the original safe for reference.
