The Japanese name for the deity Guanyin, 観音, seems as far as I can tell to be almost universally pronounced かんのん Kannon, and not かんおん Kan’on.
The origin of the name, as Wikipedia correctly gives it, is an Old or Middle Chinese translation of the Sanskrit name अवलोकितस्वर Avalokitásvara (later supplanted by अवलोकितेश्वर Avalokitéśvara) ‘he who looks down upon sound’.
The first of the two characters used in the Chinese translation, 觀 is pronounced guān /kʷan˥/ in modern Mandarin and is usually reconstructed as /kʷan/ or /kʷɑn/ in Middle Chinese. The Japanese pronunciation かん kan (earlier くゎん kwan) follows directly from this pronunciation.
The second character, 音 (yīn /(j)in˥/ in Mandarin), is reconstructed as something like /ʔiəm/ or /ʔjəm/ or /ʔi(ɪ)m/ in Middle Chinese. The Japanese Go’on pronunciation, おん on, follows fairly directly from this reconstruction, representing dialects where the /ə ~ ɪ/ was rounded and remained; while the Kan’on pronunciation, いん in, represents dialects where the initial /i ~ j/ merged with the nucleus, yielding /(j)i/. Whatever the precise form of the vowel, it definitely had a vocalic/glide initial.
Putting these two together, thus, should yield 観{かん}音{おん} kan’on, and it does seem to, rarely. Most commonly, though, it yields 観{かん}音{のん} kannon, as though 音 began with a nasal, which it obviously doesn’t.
Where does this extra nasal come from? Is it just simple 連声{れんじょう} in action?
And if so, then what about the historically almost entirely identical 漢音{かんおん}? The first character there, 漢 (Mandarin hàn /xan˦˩/) is reconstructed as Middle Chinese /xan ~ xαn/ or /han ~ hɑn/, with a non-labialised spirant rather than a labialised plosive as the initial, but with the same final (and the initials merged in Japanese anyway). So if 連声 hits 観音, it would be logical to expect it to also hit 漢音, or indeed 幹音{かんおん}, where the first character is reconstructed as /kan ~ kɑn/ and thus even closer to 観. Yet none of those seem to have gathered an extra nasal.