In Ancient Quenya, voiced stops became nasals before nasals, a sound change Tolkien described in both the Outline of Phonetic Development (OP1) from the 1930s and the Outline of Phonology (OP2) from the 1950s:

Before nasals: b, d, g were nasalized to m, n, ñ falling in with the products of p, t, k + nasal (see above). The later changes were also identical with those described for p, t, k. As noted above under (a) the medial forms were sometimes substituted, so aun from aƀn (instead of amn from abn); arn for ařn (instead of ann from adn). Thus hyarna “compact, compressed” (syadnā), after verbal stem hyar- (SYAD), beside hyanna (OP1: PE19/45).

Before nasals. In this position b, d, g, at or before the period of spirantalization, became nasalized > m, n, ñ and thus coalesced with original nasals m, n, ŋ in the same position. The later changes are identical with those described below under nasals. As was to be expected the intervocalic forms were sometimes substituted for the nasals by grammatical analogy. So aun- for aƀn- instead of amn-, arn for ařn instead of ann. Thus from √SYAD, syadnā “compressed” > hyarna “compact” for older hyanna (OP2: PE19/92).

Tolkien reaffirmed these phonetic developments in notes from the late 1960s (PE22/149-150):

In early forms b, d, g (not ʒ) were nasalized before nasals and followed A† ...

gm > ŋm > ngw ...
gn > ŋn > nn ...
dm [> nm] > nw ...
dn > nn ...
bm > mm, mb [the second development representing sporadic metathesis] ...
bn > mn

As indicated in the quote from OP2, this sound change took place before voiced stops became spirants. The nasalization of voiced stops was in some cases reformed by grammatical analogy, as indicated by the same quote. Finally, as described in the quote from OP1, in the 1930s voiceless stop + nasals had the same development, because voiceless stops were voiced before nasals. Tolkien changed his mind sometime in the late 50s or early 60s and decided that [n], [m] became [t], [w] after voiceless stops instead; see that entry for further details. Limiting ourselves to cases where the combinations were derived from a primitive voiced stop + nasal, a representative sample includes:

• √KJABA > tyamne, (archaic) past tense of Q. tyav- “to taste” (PE22/151).
• ᴹ√DUB > ᴹQ. lumna “burdensome” (Ety/DUB).
• ✶ledmē > Q. lenwe “departure” (PE17/51).
• ᴹ√LED > ᴹQ. lenna “go” (Ety/LED).
• √RIG [> rigna > riñna] > Q. -rína “crowned” (PE17/182).
• ᴹ✶sagmā > ᴹQ. sangwa “poison” (Ety/SAG).

Some of these examples are obscured by later phonetic developments whereby [nm], [ŋm] became [nw], [ŋgw] or the sporadic vocalization of [ŋ] before [n]; some of these later sound changes were also noted in the table of phonetic developments above.

Conceptual Development: There are not enough examples from the 1910s and 1920s to determine whether this sound change applied in the earliest conceptual period as well. The only relevant example I’ve found is ᴱ✶tegna > ᴱQ. tína “straight” (PE13/153, 165), which might have had a development similar to *rignā > rína above, but might also have simply spirantalized to g > ʒ [ɣ] and vocalized from there.