Samulis

Detailed overview of the cultural and political interchange around ancient Nubia, and recent archaeological discoveries. Presented by Stuart Tyson Smith, UC Santa Barbara at the Oriental Institute.

Ita vero, but that is something for translation, not a matter for a master list. For example, translating 'Yes?' into Latin might read more literally as "What is it?" (Quid est?). They have the same connotation/contextual meaning in language, but are localized to fit in the grammar and culture of each language. All good translations are not just simple word-for-word swaps, but rather translate the meaning behind the words above the actual text. Regardless, my main issue with the proposed restructuring is that it infers we will be specifically hiring or casting people for specific roles. I just don't think this will happen. Perhaps someone will perform some lines and sound like a hero or something, and we will then implement those lines as 'hero' lines, but I don't think we will be able to actively plan ahead different voices or even different sets of lines for different units beyond simple exceptions and additions. As it stands right now, I think we'll be very lucky to have at least one male & one female voice in each language within the next two years, especially the more extinct languages. I think anyone showing up on the 0AD project, wishing to record voices, just wants to see a list. They don't want to scroll through spreadsheets, dig through forums, pick a unit, have a debate, etc.. They just want to be told "here are the 20-30 lines you should record", and judging by how few voices we've had implemented and recorded so far, I think we need to simplify this process rather than make it any more complicated. So, while this restructure isn't a bad idea, it is my opinion that there is nothing broken about our current list that makes it so we must replace it, we just need more terms added and translated and clean up a few that aren't relevant, then standardize it and put it somewhere everyone can see/access, updating instructions to simplify the process and comply with our current workflow. Right now the information is there, it's just scattered and not all in agreement.

This list is a good idea, but I do not know how practical it is. Remember that there is no promise we can get different voice actors for each type of unit. People will probably continue to show up, record a bunch of lines, then leave. It's not like we have a chance to "audition" people for certain voice characteristics- whatever people choose to give to the project is what we have to work with for voices. Also keep in mind that there are many different replies currently used when a unit is selected, such as- Hello My Lord/King/Liege Yes? What is it? Similarly, when ordered to move, several different replies may occur: Yes. My Lord/King/Liege I will go The way these switch is randomly decided in the game, so it makes the voices much more enjoyable than them just saying one thing when you order the unit around. Similarly easy and practical would be to record more than one inflection of the same word, which also adds variation. I think the easiest thing for people would be just a numbered, standardized list of all the words we need recorded, which is then translated as needed into each language. I do not think we should have more than, maybe 30 lines, to keep it from being too complex. To determine that list, we first need to identify every interaction the player may have with a unit, and then determine if that interaction warrants having a separate term for it. For example, if the player orders some units to change formation, should they say something like "We will assume the formation" or is a simple "Yes, my king!" enough? Other interactions may even need several terms if they happen frequently. You have already done a good job listing most of these interactions. So, the interactions as you have listed already in your sheet: Selection Gather Food Hunt Fish Forage Farm Stone (mine) Wood (cut) Metal (mine/gather) Build Repair Garrison Building Ship (Embark) Siege Weapon/Unit Attack Move Capture Guard Patrol Special (only used by a small number of units, so they can be of less priority) Heal Trade Pack/Disassemble Unpack/Set Up Keeping in mind that Selection, Move, and Attack will account for about 50% of the gameplay experience, they should be prioritized with variation of at least 3 different lines each. The remaining lines (19) probably only need one line each. For example: Selection Yes? At your command My lord! What do you need? Move Yes! By your command I will go Attack I will attack! Charge/Advance! [localized battle-cry] For [Rome/hearth/family/the King/Kingdom, etc.]! [localized battle-cry] I will hunt I will fish I will forage/gather food I will farm I will mine stone I will cut wood I will mine/gather metal I will build I will repair I will garrison I will embark I will occupy [the siege weapon (unit)] I will capture [the building] I will guard I will patrol I will heal I will trade I will pack/prepare to move I will unpack/set up the weapon So not to make things too repetitive (although it probably makes translations harder), it might also be good to substitute the wording of a phrase or two for something more interesting or culturally/historically relevant. For example, 'I will heal' may be replaced by 'The gods will grant good health', 'I will repair' by 'It will be fixed', etc. So then someone would basically just have to take this list and translate it, then people can record straight down the list in the language(s) they choose. This makes it easy for people who just want to know what to record.

Currently on vacation recording some waves & ambiances. Still haven't had time to finish processing the horse, cow, and chicken sounds, but I should be able to work on those before the end of the year and into January.

The explicit list of sounds for when you click on a unit can be found in '\audio\voice\[language]\civ\civ_[gender]_select.xml'- Latin: <Sound>civ_female_what_is_it_1.ogg</Sound> <Sound>civ_female_hello_1.ogg</Sound> <Sound>civ_male_what_is_it_1.ogg</Sound> <Sound>civ_male_hello_1.ogg</Sound> Greek: <Sound>civ_female_what_is_it_1.ogg</Sound> <Sound>civ_female_what_is_it_2.ogg</Sound> <Sound>civ_female_what_is_it_3.ogg</Sound> <Sound>civ_female_my_lord_1.ogg</Sound> <Sound>civ_female_my_lord_2.ogg</Sound> <Sound>civ_male_yes_1.ogg</Sound> In the same folder, you will also find 'civ_[gender]_walk.xml' which is used for telling units to go places, and 'civ_[gender]_attack.xml' which is used when ordering to attack. There are also files for gathering and building, but implementation is spotty/dependent entirely on what lines were recorded in the first place. The attachment of which lines are played in which events is determined by whomever implemented the lines and is of course open to discussion and revision.

In the future, it is easier to just open the .xml files and see which <sound> files are listed. The game 'asks' the .xml files which sounds to play, so that will show you what is used where. Whether or not the .xml file is actually 'asked' by the game or not is a whole different issue of implementation that needs the input of designers/programmers to resolve, which we should address as well. Due to the haphazard nature of open development, some people don't provide all of the translations or sounds (after all, when many of these translations or recordings were provided, it is very likely there were significant differences in features/design, so they probably didn't have even had the same list we have now!). As a result, even minor changes in design as alpha testing progresses will continually result in certain voice lines being no longer used, or others being added long after the original recordings were done (e.g. capturing buildings, which has no dedicated phrase translated for any language yet- i.e. 'prendam' or 'aedificium prendam' for Latin have not been recorded or even listed). I believe things like missing words or "close enough" fits (which are very common in the current Greek audio) are okay for now until the rate of adding new big features is less. Chances are, in a year or two, there will be 2-3 more phrases added anyway. In my mind, most of the voice recordings we are doing here, even now, are probably just going to be placeholders unless the recording quality is really impressive. Most important I believe is ensuring that we have more up to date dialogue choices and translations available for voice actors so we can get all the phrases we need going forward, and none of the ones that are no longer needed. Much of that discussion needs to take place with the people responsible for designing the game and implementing the references to the .xml files, to re-examine what phrases may be no longer used and what phrases should be added. I think we would need to settle that at least before we can talk about any sort of semi-permanent implementation of voices or lots of tweaks, or else things will just continue to drift farther out between the list and the build. Part of the problem might even be that programmers have been reluctant to utilize the more specific lines because they are not available in all languages, and in some cases, not translated. Thoughts? @dmzerocold As for 'I will Work', there is no 'I will work' used in the other languages as recorded (e.g. 'laborabo'), but there is 'I will work [land/fields]' included in the Latin (agrum colam, lit. "I will cultivate the field"). I have only put 'I will work' there for now in the XML because it was just the closest fit, and whether or not you should consider re-recording is something that needs to be determined by more than any one of us.

Here are the finished files. I also re-processed some of the Greek and Latin voices to remove noise, normalize, and improve the tone a little, so all the languages should be a little more consistent with each other. These belong in the 'audio' folder within a folder called 'voice'. If you want a really quick and dirty way to try out the Persian voices, replace the Latin or Greek .ogg files with the Persian .ogg files, they should overwrite. However, there are a number of lines not used by one language compared to the other, so the result will be messy. If someone wants to go ahead and mod the Persian voices into the game, I already went through the trouble of creating .xml files for all of the lines, matching the other languages as closely as I could. Let me know if there are any issues, although I do not know how much help I can be with implementation beyond this point. voice.zip

I'll do the edits to the voice files now- rename and batch convert so they match the existing voices. Give me 1-2 hours.

Sure, I'll take care of it when I get the chance.

Thank you! It depends on when someone will implement the voices into the game. I can't make any promises, but hopefully the next major version.

Other than No2, these are all set! I will take care of the rest (cleanup, conversion, renaming, etc.). Thank you!

Hmm... no, the voice translations are kept separate from the game text translation. The voice translation are on that page on trac which you read from- that is where any voice translations should be put. If you want to post the plain text translation (non-IPA), I can at least put it on there. Yes, it is closer, but still sort of "thin" sounding compared to the No3 male. Maybe try moving the mic closer to the mouth? Your goal should be to sound consistently like the other lines you have recorded, or else people may think it is a different voice! Perfect, thank you!

If you want to write out the translations, then someone can add them to the wiki page for future voice talent, same with IPA if you are willing. Hmm... now No2 sounds like the mic is much farther from your face than the other takes. Try holding the mic the exact same place and way you held it before. You could just send the original files if you want, I should be able to extract silence from them.

Thank you! The reason things are not on the Persian list is just because no one translated them yet, I think. So, thank you for going ahead and coming up with a translation! The IPA is just a guide on how to pronounce the words the way we think they were probably said in the ancient times. It is sort of designed for people who don't speak the language or a related language, so that they can also give lines. Here is a chart, there are many others online you can look up for various languages: https://en.wiktionary.org/wiki/Wiktionary:IPA_pronunciation_key That being said, some of the existing Latin and Greek lines in the game were not recorded all that accurately. Although we would definitely appreciate as much authenticity as possible, I think if you are familiar with a modern relative of the original language, that is probably at least close enough to get a good result. My only critique is that one or two of the lines sound like they were recorded with the microphone in a different location. Take a listen to No2 and No3 for the male voice and you will hear there is a strangely thin voice sound on No2, probably the microphone was obscured. If you don't mind to re-record No2, that would be appreciated. Also, if you want me to be able to remove noise from these takes, I need you to include some 'silence'. Just hold the microphone exactly how you hold it when recording the voices, press record, then be silent and do not move, after a few seconds press stop, then send me that clip. I can use that to remove the noise from all of your takes so it sounds like it was done in a fancy studio.

The game is being constantly worked on, but the major public releases happen whenever enough changes are assembled and tested. The sounds might be implemented in the game source as soon as a few days or weeks, but it wouldn't be until the next Alpha for the sounds to reach most typical end users. Similarly, if additional sounds are added in the future, they will be rolled out as soon as the next major build comes along. I am not sure how exactly the update process works, but if there is one, the voice files will work exactly the same as any other files or assets in the game. Glad to hear! More voices, the better!

If you have a female voice talent available, sure. Each line should be done for each voice, yes. If you want to make multiple takes of each line, that is also okay too. It's better to have more takes than we need.

MUCH better! Whatever you did fixed all the issues with the audio; we can definitely work with that. Once you complete the recordings, cut the files out as before but leave a second or two of extra silence after each line. I can then go through and do any necessary denoising and alterations so they will fit with the existing game voices, then convert and organize so they can be implemented.

If you know the make and model of the mic, that might help us figure out the best way to use it. Every mic has strengths and weaknesses, it is just a matter of figuring out how to play best to the strengths and avoid the weaknesses. I would also suggest, if you are not already, using Audacity to do the recordings, as it will not do anything to the files without you telling it to, and it is very easy to configure settings for the recording. That would also minimize the risk of unwanted behavior that may occur in more "automatic" software solutions like voice recorder apps or others which might automatically modify the audio. If you have any questions about digital audio or recording, or if anything I have said is confusing or unclear, don't hesitate to ask- that's why we're here.

Hi, @dmzerocold! Thanks for sending in the voice clips! Did you apply any post-processing or conversion to these files? What equipment are you using to record this? I ask because they are at a very unusual sample rate (384 kHz). Final delivery of voice files for the game will be mono in 44.1 kHz at 16-bit, so anything higher than that is really unnecessary. Furthermore, it appears your actual physical recording equipment could only capture frequencies up to about 8 kHz and support frequencies up to 22.05 kHz, so recording at such a high sample rate is essentially useless. Take a look at this spectrograph generated from one of your takes- Across the X domain is time, and across the Y domain is frequency (labeled on the right, 0-192 kHz). The brighter the color on the spectrograph, the more sound is at that particular frequency at that point in time. You can see that above 8 kHz, the pickup of the microphone diminishes drastically, like a lowpass filter was applied (this is probably the design limit of the microphone, my guess it is a webcam mic or telephony mic). At 22.05 kHz, it essentially stops recording anything of relevance (which is to be expected for something designed to record 44.1 kHz files). All that black space at the top, from 22.05 kHz to 192 kHz, is basically wasted space- it has no relevant or valuable data in it- that's why there's no point to recording at 384 kHz; you're not even getting any reliable signal up there to record unless you're using a $1500+ calibrated test microphone. I would suggest that, if you are not doing any post-processing, then perhaps your equipment is not capable of recording the full frequency range necessary (0-20 kHz) for a video game voice over. Ideally we should be seeing a fairly consistent response in that range with nice, dark colors surrounding the voice, something like this: I've circled the same 8 kHz point (note that this file is displaying at 44.1 kHz, so the top of the graph is 22.05 kHz rather than 192 kHz as it is in yours). Notice that although, like in your voice, the bulk of the information is below about 4 kHz, there is still some valuable information up above 8 kHz, mostly what we call 'sibilance'. Without this, 's' sounds tend to appear more like 'f' sounds and also some of the character of your voice is lost, making it harder to understand and identify. If you've ever spoken to someone on the phone and thought it was someone else for a minute, that's why! Above is a clip of your recording now as it appears in the same sample rate as mine. Notice that the sibilance shown above is missing/cut off and there is less sound from 2 to 4 kHz than in my recording. My best guess is that these are issues related directly to the microphone you are using rather than the rest of your recording process. In summary, setting a higher sample rate on your computer won't enhance what isn't being captured in the first place. Unless your equipment is malfunctioning or improperly configured, resulting in that 8 kHz 'cut', you may want to consider borrowing some other equipment to see if you can get a better sound. 2nd critique- There is also a very large amount of noise in the background of the clip (all that 'pink speckles' you can see from 0-8 kHz in your recording, verus the purple and black in my recording). There are a huge number of causes of noise, including environmental sources like air conditioners, refrigerators, open windows, computer fans, and so on. In addition, there may be something in your signal flow (that is, the path the sound takes from hitting the diaphragm of the mic to being recorded on your computer as 1's and 0's) that is adding that noise. For example, if you are working with a budget consumer microphone, such as a webcam mic, it may not have adequate noise performance to record clean audio, and typically there's no real way to get around that. On the other hand, if you are recording even a very good mic but with very low gain at the preamplifier, and then add lots of gain digitally in your recording software, you will get poor noise performance there as when you record a certain 'noise floor' is "baked" into the recording and digital gain makes everything equally louder, including that noise floor. For example, here I intentionally recorded with the gain on my preamp turned down, and then used the digital gain in my recording software to make up the difference. You can see that the previously black background has now turned purple, showing how much more noise is in the recording- The lower the noise we can record, the better the end result will be! So, a few tips for reducing noise in general- Turn off anything that makes noise in your recording environment. Find a location which is enclosed and away from any sources of sound like busy streets. Record as close to your mic as you can, preferably 4-8 inches (10-20 cm). The closer you get, the louder your voice will sound at the mic and thus the quieter noise will appear. There are some downsides to recording close to a mic, so try holding it in different positions- above your mouth pointed slightly down towards it, below your mouth pointed slightly up towards it, and so on, to find what sounds best. Record at 44.1 kHz 16-bit in Mono (although you can also record at 24-bit if you want, this really won't change anything at this sort of noise level). Make sure any analog amplification is turned up before using any digital gain. For example, if there is a gain knob on the microphone or the preamplifier itself, use that before using the slider in Windows or your recording application. If your microphone can be configured to different frequency ranges (for example, some webcam or USB mics), make sure it is set to 44.1 kHz too. If you still have similarly noisy results, perhaps consider borrowing a different microphone from a friend. Even something like a Blue Snowball should give significantly cleaner and better sounding results.

Thanks Omri! That's the idea, indeed. Something non-specific but different (e.g. different shields made of wood) can be encoded in the 'type' number, but for anything that is very likely to have a specific, unique end use that should be made clear, such as footsteps on different material, will have a clear, verbose name as you suggest. shield_hit_wood_1_... shield_hit_wood_2_...

Thanks guys! If anyone has any feedback on the naming format, please feel free to share it. Otherwise, we should be good to publish (I assume I need to set up an account on the Wiki to edit?). Timeline for getting the horse/cow/chicken samples up is probably 1-2 weeks. Unsurprisingly there's a lot of background noise (birds, crickets, cars, etc.) that has to be cleaned out for the sounds to be usable.

Hi all, For the past few months, I've been working on taking stock of the existing sound effects in the game and recording some new material. I currently have generated a catalog of about 10 GB of raw recordings to go through, and I've been making slow but steady headway editing this material to get down to final sound effects that can go into the game (cut, process, clean, convert). That probably sounds like a lot, but after all the editing and converting from 48kHz/24-bit .wav to the .ogg vorbis format, the end result is maybe going to be 150-300 MB total. This first set will mostly be ambiance sound effects, spurred on by @stanislas69. I currently have 83 sound files in this category, which I've temporarily placed in a public dropbox folder (click the link below, then click 'download' in the top right corner of the page to download it all in one go). All of these sounds are from my own recordings, most of which were specifically recorded for this project. https://www.dropbox.com/sh/7m5tg9ln1xjmlnw/AAD2hD-Aorl1YkYXPqcgJpUea?dl=0 I don't have XML files generated yet, but the naming convention (a subtle but clearer evolution upon the original convention) should be more or less finalized, based on one discussed in my application thread, implemented here for feedback with revisions proposed by GunChleoc- With the naming convention, we basically inserted two extra underscores, so the values are [name]_[type]_[intensity]_[variation]. Type is like the difference between two different birds or microphones, while variation is two different takes done by the same bird or microphone setup. Within a type and intensity you are essentially guaranteed a certain degree of similarity. Intensity is how much of whatever is going on, for example, some calm, sparse waves versus stronger, more stormy waves, or one person walking versus 20 people walking. The thunder sounds are a good example of the difference. I think Implementation would probably be to make one .xml file for each name_type (e.g. bird_1.xml, bird_2.xml, etc.). I have temporarily organized the files into six categories under 'Ambiance'- Forest Plains Swamp Thunder Water Coast Flowing (e.g. streams, rivers, waterfalls- encoded in the intensity parameter of the name) Rain Wind I would also like to reiterate that this is just the beginning, and that over time more files and likely categories will be added (e.g. 'desert', 'Mediterranean', etc.). With that in mind, I am looking for people interested in recording ambiances for the game, specifically those living in Europe, Africa, or the Middle East who have access to quiet, un-populated areas (e.g. natural parks, rural areas, etc.). As I live and work in New England, I do not exactly have an easy way to record all the different climates of the world, as much as I would enjoy a vacation. You don't need much in terms of equipment, just a little hand-held device like a Zoom H1 with a windscreen would do just fine (a cell phone's built-in mic isn't quite good enough). Find a quiet spot (edit: by that I mean as far away from any sound of human activity as possible, not really 'quiet'), move the dial until the sound is between -24 and -12 on the display, press record while holding it (or mounting it on a tripod/stand) towards the sound source (and away from any unwanted sound), remain very quiet and still for a minute or two, then stop the recording. I can take care of the rest! Anyway, let me know what you all think, if you have any feedback, questions, or concerns, it is all welcome (especially before I edit too many of these!). A quick preview of what I have coming up next to edit. I plan to make new tasks for each of these sets for clarity, but continue using the same dropbox- Horse, chicken, and cow sounds (walking, vocalizing, eating) Loads of foley, movement, fire, and weapon sounds

Oooh, one step ahead of you then; recorded some horse grazing sounds.

Will do! I am about to go on a short "workcation" for a week, but when I get back, I'll post up some of the sounds so far. I still need to do clean-up, like removing when a bird chirps in the horse walking sound and so on. Should I post it in this 'Official Tasks' forum or the 'Unofficial Tasks' forum since it doesn't have a related Trac/Github fork at the moment and would just be some loose audio files?

If you keep recording fairly close to the mic, there's no point in getting acoustic panels. Besides, you won't really notice much of a change with those products unless you use a lot of them and are very strategic in their placement- for example, just sticking six of them on a wall won't really do all that much. On the other hand, if you have a hard floor (tile/stone/wood), getting something like a big area rug would help quite a bit and can also be very cheap. Generally the thicker/more fluffy the rug, the better it works. Definitely the biggest area for improvement would be trying to get rid of the hum/buzz. If your wireless receiver and transmitter have different channels, you could see if that helps too. I'm currently a bit busy at the moment with a bunch of different projects, but I am continuing work on the new sounds as I can. I recorded an additional 3 GB last week, including some "normal" horses, cattle, and a few other things. I also finished sorting out my earlier recordings and even started processing some of them (already have a bunch of draft horse sounds that I'd like to put up soon). I figure that's a topic for another thread, though.