I would normally close this as a duplicate, but you've linked to the question it's a duplicate of. What are you looking for that's not found there?

Unlike modern languages, where we can simply listen to native speakers and describe the sounds they use in detail, the very nature of reconstructed proto-languages means that we can say very, very little about how things actually sounded. We can make educated guesses at what sound values specific morphophonemic units had, which is what the answer to the linked question does, but we can do little more. Once you’ve decided which sound you want to associate with each reconstructed unit, an ‘IPA’ representation is a simple matter of substitution – so what exactly are you looking for here?

Hi, thanks for the comments! I am looking for some source that has (in the best case scenario) a list of best guesses for how each word is pronounced, sort of like a dictionary with phonetic information attached to each word. Otherwise, a generic list of instructions on how I can myself try to deduce possible pronunciations for a substantial number of words might also help, so that i can try to construct something like that dictionary myself for my use

in that case this seems to me to be an exact duplicate of the other question

Does this answer your question? Proto-Indo-European (PIE) words with IPA

In that case, the only thing you need is a table like the one in the answer in the linked question (you decide which of the proposed correspondences you prefer). That has IPA values for each reconstructed PIE unit and you can simply take any PIE word you find and substitute the reconstructed form with the IPA entities in your table. If you decide you want use /x/ for *h₂, /kʲ/ for *k̑ (*ḱ, *k̂) and /ε/ for *e, then you can quite mechanically transcribe *h₂ék̑mōn ‘rock’ into IPA as /ˈxεkʲmoːn/. If you want to include laryngeal colouring, it’s equally mechanically /ˈxakʲmoːn/.

@JanusBahsJacquet as laryngeal colouring (at least a-colouring of *h2) is attested in all branches (including Anatolian), it ought to be reconstructed as already occurring at the phonetic level in PIE (although it was certainly not phonemic yet)

@Tristan Probably not phonemic, no, but you can argue for including or excluding it from IPA transcriptions, depending on where on the phonemic/phonetic spectrum you’re aiming for. My point about the colouring was simply that it doesn’t really change anything – the colouring is automatic, so the transcription remains mechanical.

@JanusBahsJacquet transcribing *k̑ as /kʲ/ makes many words absolutely impossible to pronounce, if it means the same as Russian soft k.

@Anixx Does it? Why? Any examples? There are lots of PIE words that I find quite hard to wrap my tongue around, but the difficulties for me mostly lie with consecutive laryngeals – I’ve never had any particular trouble with *k̑.

@JanusBahsJacquet take for instance root *ḱh₂d-. Palatalized velar plosive followed by non-palatalized velar fricative? Are you serious?

@Anixx I don’t find that particularly difficult to pronounce… and at any rate, in zero-grade sequence like that, the laryngeal would be vocalised, so it would be *k̑ə₂d-. The most difficult part for me in pronouncing /kʲxd-/ or /kʲₔxd/ is the mixed voicing in /xd/, which almost inevitably becomes /xt/ when I say it. But such clusters are well-attested in modern languages; it’s just me finding it difficult.

@JanusBahsJacquet I cannot imagine how one can pronounce a soft consonant followed by a hard consonant. It is impossible in Russian and I cannot pronounce this. This concerns clusters ḱr-, ḱw-, ḱs- etc. Also, soft consonant necessarily makes the following vowel front, so again, this could not pass without affecting vowels. How one is supposed to pronounce *swéḱsdḱomt?

@Anixx You find it impossible because it’s impossible in your native language. That doesn’t mean it’s impossible elsewhere. And no, palatalised consonants do not necessarily make the following vowel front; it’s perfectly commonplace to have non-front vowels after palatalised consonants, e.g., English gnaw /ɲɔː/, French oignon /ɔˈɲɔ̃/, Irish ceol /kʲoːl/, etc. Again, I find [ˈswekʲs(t)kʲomt] relatively easy to pronounce (though with voicing assimilation – if the /d/ has to be voiced, that’s difficult for me).

@JanusBahsJacquet I have just tried to listen to your examples in Google speech synthesizer, and in "gnaw" I do not hear anything like a soft consonant in Russian, and in oignon I hear front vowel ö.

@JanusBahsJacquet "gnaw" has a palatal nasal for you? I've never heard such a pronunciation, or seen it suggested for any stage of English (only ever undergoing a unitary /gn/ > /n/ shift). Otherwise your comment is good. Russian itself has plenty of examples of paltalised consonants followed by non-front vowels, as seen when Я, Ё, & Ю follow most consonants

@anixx whilst a restriction that clusters be alike for palatalisation is common (it occurs independently in e.g. Russian and Irish), this is still a restriction particular to those languages, and not a general restriction in the ability of the human vocal apparatus

there are plenty of reasons to doubt the identification of ḱ with /kʲ/, but they mostly relate to markedness and relative frequency and not the effects your suggesting (in particular that *ḱ is overwhelmingly more common than *k and to a much lesser extent *kʷ, and likewise for the voiced and voiced aspirates; and so the palatovelars ought to be *less marked than them, which would suggest it ought to be /k/)

@Tristan we are talking about palatalized consonants(soft in Russian terminology), not palatal. Я, Ё, & Ю in Russian are front vowels.

@Tristan I agree that ḱ should be a plain velar and k should be a "super-hard" consonant (like q in Turkic).

@Tristan Actually it doesn’t, no… I originally wrote gnu, but then wanted an example with an unambiguously non-front vowel (/uː/ often phonetically starting with something closer to [ʉ] or [y]), and gnaw popped into my mind. I have definitely heard it with a palatalised nasal in old-timey theatrical speech, but I wouldn’t pronounce it like that myself – should have gone for gnocchi instead.

@Anixx If you’re hearing any kind of front vowel in oignon, that must be due to influence from Russian in your perception. It is absolutely not a front vowel by any means. The same is true in Italian gnocchi, Spanish sueño, etc.

@JanusBahsJacquet I have just tried to listen to it in Google sheech synthesizer. I definitely hear front vowels there, in both cases, what is denoted with letter ё in Russian.

@Anixx Well, that’s on you, because it’s not a front vowel, and the Google speech synthesizer is not pronouncing it as one either (I just checked). You’re simply wrong. The /ɔ̃/ vowels in oignon, mont and ton are the same rounded nasal back vowel. Russian ⟨ë⟩ (= /o/ after a soft consonant) is a centralised rounded [ɵ] which does not exist at all in French (and which, incidentally, is not a front vowel, but a central vowel); non-nasal /ɔ/ is sometimes centralised to something like [ɞ] in some French dialects, especially before /ʁ/, but the nasal vowel is not.

@Anixx - »I cannot imagine how one can pronounce a soft consonant followed by a hard consonant. It is impossible in Russian and I cannot pronounce this.« — But in fact Russian is full of the most common words with a “soft” + “hard” consonant clusters, e.g.: только, больше, синька, писька, бочка, лайка, etc.

@Anixx Ё, & Ю are back vowels /o/, & /u/ (at least when stressed), but mark palatalisation of a preceding consonant (or, if not immediately following a consonant, the insertion of a /j/ before the vowel). Я is prototypically /a/ (so central) when stressed, but can front to front /æ/ in some positions so was less good an example. From a linguistic pov, palatalisation is a property of the consonant, but it sounds like, under the influence of Russian orthography you're perceiving it as a property of the vowel (and then calling it "frontness", which is another property altogether)

@JanusBahsJacquet honestly trying to use an English word at all is probably a lost cause, as /ɲ/ is a marginal loanphoneme at best (and ime foreign [ɲ] is typically approximated with the cluster /nj/ instead of with a loan phoneme). I am surprised you've heard it even in theatrical speech for gnaw though seeing as that's a native word

I should specify, when discussing the pronunciation of Я, Ё, & Ю I am specifically talking about the vowel part. Due to the nature of Russian orthography, they mark both a vowel and the quality of a preceding consonant (or iotation if there is no immediately preceding orthographic consonant), but from a phonological point of view, that quality belongs to the consonant and not to the vowel

@Tristan I do not know what is back and front vowels then in your terminology. In my e, ё, ю, я, и are front vowels, э, у, а, ы are back vowels (but э is front vowel word-initially). This is so simple and straightforward. ё is the same as German ö and ю is the same as ü (when they are not iotized).

@YellowSky you are right, so not all consonant clusters of this form are difficult.

@Anixx I don’t know what kind of strange Russian or German you speak, but Russian ⟨ë, ю⟩ and German ⟨ö, ü⟩ are absolutely not in any way, under any circumstances, even remotely similar. German ⟨ö⟩ is either /œ/ or /øː/ and ⟨ü⟩ is either /ʏ/ or /yː/, none of which exists in Russian at all. Tristan’s terminology is that of standard linguistics; yours appears to be one that describes the vowels in the Russian orthography, but it’s certainly not one used in phonetics in general.

@Tristan "From a linguistic pov, palatalisation is a property of the consonant" - no. Actually, one can choose whether to count it as a property of vowel or consonant. In Russian it is called a property of consonant becaus soft/hard consonant difference is phonemic word-finally, while front/back vowel is not phonemic word-initially in Russian. So, it is natural to count it a property of consonant. In the middle of the word, soft consonant always follows with a front vowel.

@JanusBahsJacquet I do not know what these IPA symbols even mean. They seem to be English-centric. Russian ⟨ë, ю⟩ are exactly German ⟨ö, ü⟩ except the Russian ones are iotised in certain positions. Schröder is Шрёдер in Russian, Münz is Мюнц, etc, and it accurately reflects German pronunciation. Russian only cannot reflect in spelling words starting with ö, or ü because ё and ю get iotized word-initially.

@Anixx no, that is the normal transliteration of German into Russian, but it does not accurately reflect German pronunciation. Your perspective is being entirely clouded by Russian orthography. I strongly suggest you read up on phonetics because your current position is incorrect

@Tristan so you claim I should learn IPA orthography? German words are perfectly transliterated this way for a reason. No-one complains it is a bad convention. I have just read the Wikipedia article about Russian phonology, and it is a total mess. For instance, it claims that я in пять and пятка is read differently and in the first case it is front vowel, but not in the second case. This is an absolutely ridiculous claim, they are pronounced exactly the same way.

@Tristan I wonder, who invented all these ridiculous claims? Or that ё in тётя and ы in тыква are both mid-vowels, neither back, nor front, when in fact ё inpronounced with lips and front part of tongue while ы is pronounced by the back part of the tongue. It seems those who invented IPA could not agree what their characters represent. It has the seal of the mess with vowels characteristic of English language since the Medieval times.

@Tristan soft and hard vowels in Russian are easily distinguishable by the native speakers without the preceding consonants by the way, so the claim that the vowels я and а are read the same in пятка and пат is also ridiculous. Russian speakers can produce the я sound isolated and prolong it as they wish, just as a sound. They are considered not differentiated phonemically but they can be easily produced and differentiated by sound. And it is only a convention to consider the phonemic difference to fall on the consonants and not on vowels

your claims are universally rejected by linguists, continuing this conversation is unlikely to be productive