The Tampax Tapes were recorded on a bugged phone.
The Sydney Morning Herald reports:
"On the night in question, it is widely acknowledged that Charles was staying at Eaton Hall in Cheshire, the 4400-hectare family estate of one of his closest friends and one of England's wealthiest landowners, the Duke of Westminster.
The minutes from a top secret security meeting held in Whitehall in April 1993 revealed that Lord Fellowes, then the Queen's private secretary, said: ''Evidence had been found that the fixed telephone lines had been tampered with [at Eaton Hall]. It was almost certain that this was the location where the Prince of Wales had been staying on the night of the alleged conversation between him and Mrs Parker Bowles.''
Then cabinet secretary Sir Robin Butler was so alarmed by the phone hacking that he asked the meeting whether all royal premises should be ''swept'' for bugs."
The Wikipedia entry about the Squidgygate tapes is also interesting. In 1993, The Sunday Times published the findings of an analysis of the "Squidgygate" tape, commissioned from Corby-based surveillance specialists Audiotel International.
"Audiotel concluded that the presence of data bursts on the tape was suspicious. Data bursts ("pips" at intervals of approximately 10 seconds, containing information for billing purposes) would normally be filtered out at the exchange before Cellnet transmission. That these "pips" were present at all was therefore anomalous, but they were also too fast, too loud, and exhibited a "low-frequency [audio] 'shadow'," implying "some kind of doctoring of the tape," said Audiotel's managing director, Andrew Martin, in his firm's report. "The balance of probability suggests something irregular about the recording which may indicate a rebroadcasting of the conversation some time after the conversation took place."[9]
"Within a week of the Times's announcement, a further independent analysis was carried out for the same newspaper by John Nelson of Crew Green Consulting, with assistance from Martin Colloms, audio analyst for Sony International. Their analysis demonstrated convincingly that the conversation could not have been recorded by a scanning receiver in the manner claimed by Mr Reenan. Amongst several relevant factors, there was a 50 hertz hum in the background of the "Squidgygate" conversation together with components in the recorded speech with frequencies in excess of 4 kHz. Neither could have passed through the filters of Mr Reenan's Icom receiver or indeed have been transmitted by the cellular telephone system. The 50 Hz hum was consistent with the effect of attempting to record a telephone conversation via a direct tap on a landline."
Turns out Diana had good reason to be "paranoid".